net: Improve netns handling in RTNL and ip_tunnel
Subsystems (3)
cover net: Improve netns handling in RTNL and ip_tunnel
This patch series includes some netns-related improvements and fixes for RTNL and ip_tunnel, to make link creation more intuitive:
- Creating link in another net namespace doesn't conflict with link names in current one.
- Add a flag in rtnl_ops, to avoid netns change when link-netns is present if possible.
- When creating ip tunnel (e.g. GRE) in another netns, use current as link-netns if not specified explicitly.
So that
# modprobe ip_gre netns_atomic=1 # ip link add netns ns1 link-netns ns2 tun0 type gre ...
will create tun0 in ns1, rather than create it in ns2 and move to ns1. And don't conflict with another interface named "tun0" in current netns.
---
Xiao Liang (5):
rtnetlink: Lookup device in target netns when creating link
rtnetlink: Add netns_atomic flag in rtnl_link_ops
net: ip_tunnel: Build flow in underlay net namespace
net: ip_tunnel: Add source netns support for newlink
net: ip_gre: Add netns_atomic module parameter
include/net/ip_tunnels.h | 3 +++ include/net/rtnetlink.h | 3 +++ net/core/rtnetlink.c | 15 ++++++++++----- net/ipv4/ip_gre.c | 15 +++++++++++++-- net/ipv4/ip_tunnel.c | 27 ++++++++++++++++++--------- 5 files changed, 47 insertions(+), 16 deletions(-)
Signature · 1 line
2.47.0
PATCH 1/5 rtnetlink: Lookup device in target netns when creating link
When creating link, lookup for existing device in target net namespace instead of current one.
For example, two links created by:
# ip link add dummy1 type dummy # ip link add netns ns1 dummy1 type dummy
should have no conflict since they are in different namespaces.
--- net/core/rtnetlink.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)
net/core/rtnetlink.c +6 -2
@@ -3733,20 +3733,24 @@ static int __rtnl_newlink(struct sk_buff *skb, struct nlmsghdr *nlh, {struct nlattr **const tb = tbs->tb;struct net *net =sock_net (skb->sk); +struct net *device_net;struct net_device *dev;struct ifinfomsg *ifm;bool link_specified; +/* When creating, lookup for existing device in target net namespace */ + device_net = nlh->nlmsg_flags &NLM_F_CREATE ? tgt_net : net; + ifm =nlmsg_data (nlh);if (ifm->ifi_index >0 ) { link_specified = true; - dev =__dev_get_by_index (net, ifm->ifi_index); + dev =__dev_get_by_index (device_net, ifm->ifi_index); }else if (ifm->ifi_index <0 ) {NL_SET_ERR_MSG (extack,"ifindex can't be negative" );return -EINVAL ; }else if (tb[IFLA_IFNAME ] || tb[IFLA_ALT_IFNAME ]) { link_specified = true; - dev =rtnl_dev_get (net, tb); + dev =rtnl_dev_get (device_net, tb); }else { link_specified = false; dev =NULL ;
Signature · 1 line
2.47.0
Re: [PATCH net-next 1/5] rtnetlink: Lookup device in target netns when creating link
From: Xiao Liang <[email protected]>
Date: Wed, 23 Oct 2024 10:31:42 +0800
Quoted 19 lines
When creating link, lookup for existing device in target net namespace instead of current one.
For example, two links created by:# ip link add dummy1 type dummy # ip link add netns ns1 dummy1 type dummyshould have no conflict since they are in different namespaces.
Signed-off-by Xiao Liang <[email protected]>--- net/core/rtnetlink.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)net/core/rtnetlink.c +0
-0
net/core/rtnetlink.c static int __rtnl_newlink(struct sk_buff *skb, struct nlmsghdr *nlh, · line 3742
Quoted 10 lines
net/core/rtnetlink.c +3
-0@@ -3733,20 +3733,24 @@ static int __rtnl_newlink(struct sk_buff *skb, struct nlmsghdr *nlh, {struct nlattr **const tb = tbs->tb;struct net *net =sock_net (skb->sk); +struct net *device_net;struct net_device *dev;struct ifinfomsg *ifm;bool link_specified; +/* When creating, lookup for existing device in target net namespace */ + device_net = nlh->nlmsg_flags &NLM_F_CREATE ? tgt_net : net;
Technically, this changes uAPI behaviour.
Let's say a user wants to
1) move the device X in the current netns to another if exists, otherwise
2) create a new device X in the target netns
This can be achieved by setting NLM_F_CREATE and IFLA_NET_NS_PID, IFLA_NET_NS_FD, or IFLA_TARGET_NETNSID.
But with this change, the device X in the current netns will not be moved, and a new device X is created in the target netns.
Quoted 18 lines
net/core/rtnetlink.c +3
-2+ ifm =nlmsg_data (nlh);if (ifm->ifi_index >0 ) { link_specified = true; - dev =__dev_get_by_index (net, ifm->ifi_index); + dev =__dev_get_by_index (device_net, ifm->ifi_index); }else if (ifm->ifi_index <0 ) {NL_SET_ERR_MSG (extack,"ifindex can't be negative" );return -EINVAL ; }else if (tb[IFLA_IFNAME ] || tb[IFLA_ALT_IFNAME ]) { link_specified = true; - dev =rtnl_dev_get (net, tb); + dev =rtnl_dev_get (device_net, tb); }else { link_specified = false; dev =NULL ;Signature · 1 line
2.47.0
Re: [PATCH net-next 1/5] rtnetlink: Lookup device in target netns when creating link
Quoted 19 lines · Kuniyuki Iwashima
Quoted 4 lines
net/core/rtnetlink.c static int __rtnl_newlink(struct sk_buff *skb, struct nlmsghdr *nlh, · line 3742
Quoted 10 lines
net/core/rtnetlink.c +3
-0@@ -3733,20 +3733,24 @@ static int __rtnl_newlink(struct sk_buff *skb, struct nlmsghdr *nlh, {struct nlattr **const tb = tbs->tb;struct net *net =sock_net (skb->sk); +struct net *device_net;struct net_device *dev;struct ifinfomsg *ifm;bool link_specified; +/* When creating, lookup for existing device in target net namespace */ + device_net = nlh->nlmsg_flags &NLM_F_CREATE ? tgt_net : net;Technically, this changes uAPI behaviour.
Let's say a user wants to
Quoted 8 lines
1) move the device X in the current netns to another if exists, otherwise
2) create a new device X in the target netnsThis can be achieved by setting
NLM_F_CREATEandIFLA_NET_NS_PID,IFLA_NET_NS_FD, orIFLA_TARGET_NETNSID.But with this change, the device X in the current netns will not be moved, and a new device X is created in the target netns.
You're right, what about testing for NLM_F_EXCL aslo?
PATCH 2/5 rtnetlink: Add netns_atomic flag in rtnl_link_ops
In reply to Xiao LiangCurrently these two steps are needed to create a net device with IFLA_LINK_NETNSID attr:
1. create and setup the netdev in the link netns with rtnl_create_link()
2. move it to the target netns with dev_change_net_namespace()
This has some side effects, including extra ifindex allocation, ifname validation and link notifications in link netns.
Add a netns_atomic flag, that if set to true, devices will be created in the target netns directly.
--- include/net/rtnetlink.h | 3 +++ net/core/rtnetlink.c | 7 ++++--- 2 files changed, 7 insertions(+), 3 deletions(-)
include/net/rtnetlink.h +3 -0
@@ -74,6 +74,8 @@ static inline int rtnl_msg_family(const struct nlmsghdr *nlh)* @srcu: Used internally * @kind: Identifier * @netns_refund: Physical device, move to init_net on netns exit +* @netns_atomic: Device can be created in target netns even when +* link-netns is different, avoiding netns change. * @maxtype: Highest device specific netlink attribute number * @policy: Netlink policy for device specific attribute validation * @validate: Optional validation function for netlink/changelink parameters
@@ -115,6 +117,7 @@ struct rtnl_link_ops {void (*setup)(struct net_device *dev);bool netns_refund; +bool netns_atomic;unsigned int maxtype;const struct nla_policy *policy; int (*validate)(struct nlattr *tb[],
net/core/rtnetlink.c +4 -3
@@ -3679,8 +3679,9 @@ static int rtnl_newlink_create(struct sk_buff *skb, struct ifinfomsg *ifm, name_assign_type =NET_NAME_ENUM ; } - dev =rtnl_create_link (link_net ? : tgt_net, ifname, - name_assign_type, ops, tb, extack); + dev =rtnl_create_link (!link_net || ops->netns_atomic ? + tgt_net : link_net, ifname, name_assign_type, + ops, tb, extack);if (IS_ERR (dev)) { err =PTR_ERR (dev); goto out;
@@ -3700,7 +3701,7 @@ static int rtnl_newlink_create(struct sk_buff *skb, struct ifinfomsg *ifm, err =rtnl_configure_link (dev, ifm, portid, nlh);if (err <0 ) goto out_unregister; -if (link_net) { +if (link_net && !ops->netns_atomic) { err =dev_change_net_namespace (dev, tgt_net, ifname);if (err <0 ) goto out_unregister;
Signature · 1 line
2.47.0
Re: [PATCH net-next 2/5] rtnetlink: Add netns_atomic flag in rtnl_link_ops
From: Xiao Liang <[email protected]>
Date: Wed, 23 Oct 2024 10:31:43 +0800
Quoted 6 lines
Currently these two steps are needed to create a net device with
IFLA_LINK_NETNSIDattr:1. create and setup the netdev in the link netns with
rtnl_create_link()
2. move it to the target netns withdev_change_net_namespace()
IIRC, this is to send the notification in the link netns.
Quoted 62 lines
This has some side effects, including extra ifindex allocation, ifname validation and link notifications in link netns.
Add a
netns_atomicflag, that if set to true, devices will be created in the target netns directly.Signed-off-by Xiao Liang <[email protected]>--- include/net/rtnetlink.h | 3 +++ net/core/rtnetlink.c | 7 ++++--- 2 files changed, 7 insertions(+), 3 deletions(-)include/net/rtnetlink.h +3
-0@@ -74,6 +74,8 @@ static inline int rtnl_msg_family(const struct nlmsghdr *nlh)* @srcu: Used internally * @kind: Identifier * @netns_refund: Physical device, move to init_net on netns exit +* @netns_atomic: Device can be created in target netns even when +* link-netns is different, avoiding netns change. * @maxtype: Highest device specific netlink attribute number * @policy: Netlink policy for device specific attribute validation * @validate: Optional validation function for netlink/changelink parameters @@ -115,6 +117,7 @@ struct rtnl_link_ops {void (*setup)(struct net_device *dev);bool netns_refund; +bool netns_atomic;unsigned int maxtype;const struct nla_policy *policy; int (*validate)(struct nlattr *tb[],net/core/rtnetlink.c +4
-3@@ -3679,8 +3679,9 @@ static int rtnl_newlink_create(struct sk_buff *skb, struct ifinfomsg *ifm, name_assign_type =NET_NAME_ENUM ; } - dev =rtnl_create_link (link_net ? : tgt_net, ifname, - name_assign_type, ops, tb, extack); + dev =rtnl_create_link (!link_net || ops->netns_atomic ? + tgt_net : link_net, ifname, name_assign_type, + ops, tb, extack);if (IS_ERR (dev)) { err =PTR_ERR (dev); goto out;@@ -3700,7 +3701,7 @@ static int rtnl_newlink_create(struct sk_buff *skb, struct ifinfomsg *ifm, err =rtnl_configure_link (dev, ifm, portid, nlh);if (err <0 ) goto out_unregister; -if (link_net) { +if (link_net && !ops->netns_atomic) { err =dev_change_net_namespace (dev, tgt_net, ifname);if (err <0 ) goto out_unregister;Signature · 1 line
2.47.0
Re: [PATCH net-next 2/5] rtnetlink: Add netns_atomic flag in rtnl_link_ops
Quoted 12 lines · Kuniyuki Iwashima
From: Xiao Liang <[email protected]>
Date: Wed, 23 Oct 2024 10:31:43 +0800Quoted 6 lines
Currently these two steps are needed to create a net device with
IFLA_LINK_NETNSIDattr:1. create and setup the netdev in the link netns with
rtnl_create_link()
2. move it to the target netns withdev_change_net_namespace()IIRC, this is to send the notification in the link netns.
Yes. This patch changes this behavior only when the new flag is set.
I doubt if it's really necessary to send link create/delete notifications to link-netns. Also the current behavior is somewhat inconsistent, say
1) ip link add netns n1 link-netns n2 link eth0 mac0 type macvlan
2) ip -n n2 link add netns n1 link eth0 mac0 type macvlan
Intuitively, the two commands are equivalent. But notification is sent only for 1.
PATCH 3/5 net: ip_tunnel: Build flow in underlay net namespace
In reply to Xiao LiangBuild IPv4 flow in underlay net namespace, where encapsulated packets are routed.
--- net/ipv4/ip_tunnel.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
net/ipv4/ip_tunnel.c +3 -3
@@ -294,7 +294,7 @@ static int ip_tunnel_bind_dev(struct net_device *dev)ip_tunnel_init_flow (&fl4, iph->protocol, iph->daddr, iph->saddr, tunnel->parms.o_key, - iph->tos &INET_DSCP_MASK ,dev_net (dev), + iph->tos &INET_DSCP_MASK , tunnel->net, tunnel->parms.link, tunnel->fwmark,0 ,0 ); rt =ip_route_output_key (tunnel->net, &fl4);
@@ -611,7 +611,7 @@ void ip_md_tunnel_xmit(struct sk_buff *skb, struct net_device *dev, }ip_tunnel_init_flow (&fl4, proto, key->u.ipv4.dst, key->u.ipv4.src,tunnel_id_to_key32 (key->tun_id), - tos &INET_DSCP_MASK ,dev_net (dev),0 , skb->mark, + tos &INET_DSCP_MASK , tunnel->net,0 , skb->mark,skb_get_hash (skb), key->flow_flags);if (!tunnel_hlen)
@@ -774,7 +774,7 @@ void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,ip_tunnel_init_flow (&fl4, protocol, dst, tnl_params->saddr, tunnel->parms.o_key, tos &INET_DSCP_MASK , -dev_net (dev),READ_ONCE (tunnel->parms.link), + tunnel->net,READ_ONCE (tunnel->parms.link), tunnel->fwmark,skb_get_hash (skb),0 );if (ip_tunnel_encap (skb, &tunnel->encap, &protocol, &fl4) <0 )
Signature · 1 line
2.47.0
PATCH 4/5 net: ip_tunnel: Add source netns support for newlink
In reply to Xiao LiangAdd ip_tunnel_newlink_net() that accepts src_net parameter, which is passed from newlink() of RTNL ops, and use it as tunnel source netns.
--- include/net/ip_tunnels.h | 3 +++ net/ipv4/ip_tunnel.c | 21 +++++++++++++++------ 2 files changed, 18 insertions(+), 6 deletions(-)
include/net/ip_tunnels.h +3 -0
@@ -408,6 +408,9 @@ int ip_tunnel_changelink(struct net_device *dev, struct nlattr *tb[], struct ip_tunnel_parm_kern *p, __u32 fwmark);int ip_tunnel_newlink (struct net_device *dev,struct nlattr *tb[],struct ip_tunnel_parm_kern *p,__u32 fwmark); +int ip_tunnel_newlink_net (struct net *src_net,struct net_device *dev, +struct nlattr *tb[],struct ip_tunnel_parm_kern *p, +__u32 fwmark);void ip_tunnel_setup (struct net_device *dev,unsigned int net_id); bool ip_tunnel_netlink_encap_parms(struct nlattr *data[],
net/ipv4/ip_tunnel.c +15 -6
@@ -1213,17 +1213,17 @@ void ip_tunnel_delete_nets(struct list_head *net_list, unsigned int id, }EXPORT_SYMBOL_GPL (ip_tunnel_delete_nets); -int ip_tunnel_newlink (struct net_device *dev,struct nlattr *tb[], -struct ip_tunnel_parm_kern *p,__u32 fwmark) +int ip_tunnel_newlink_net (struct net *src_net,struct net_device *dev, +struct nlattr *tb[],struct ip_tunnel_parm_kern *p, +__u32 fwmark) {struct ip_tunnel *nt; -struct net *net =dev_net (dev);struct ip_tunnel_net *itn;int mtu;int err; nt =netdev_priv (dev); - itn =net_generic (net, nt->ip_tnl_net_id); + itn =net_generic (src_net, nt->ip_tnl_net_id);if (nt->collect_md) {if (rtnl_dereference (itn->collect_md_tun))
@@ -1233,7 +1233,7 @@ int ip_tunnel_newlink(struct net_device *dev, struct nlattr *tb[],return -EEXIST ; } - nt->net = net; + nt->net = src_net; nt->parms = *p; nt->fwmark = fwmark; err =register_netdevice (dev);
@@ -1265,6 +1265,13 @@ int ip_tunnel_newlink(struct net_device *dev, struct nlattr *tb[], err_register_netdevice:return err; } +EXPORT_SYMBOL_GPL (ip_tunnel_newlink_net); + +int ip_tunnel_newlink (struct net_device *dev,struct nlattr *tb[], +struct ip_tunnel_parm_kern *p,__u32 fwmark) +{ +return ip_tunnel_newlink_net (dev_net (dev), dev, tb, p, fwmark); +}EXPORT_SYMBOL_GPL (ip_tunnel_newlink); int ip_tunnel_changelink(struct net_device *dev, struct nlattr *tb[],
@@ -1326,7 +1333,9 @@ int ip_tunnel_init(struct net_device *dev) } tunnel->dev = dev; - tunnel->net =dev_net (dev); +if (!tunnel->net) + tunnel->net =dev_net (dev); +strscpy (tunnel->parms.name, dev->name); iph->version =4 ; iph->ihl =5 ;
Signature · 1 line
2.47.0
PATCH 5/5 net: ip_gre: Add netns_atomic module parameter
In reply to Xiao LiangIf set to true, create device in target netns (when different from link-netns) without netns change, and use current netns as link-netns if not specified explicitly.
--- net/ipv4/ip_gre.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-)
net/ipv4/ip_gre.c +13 -2
@@ -107,6 +107,11 @@ static bool log_ecn_error = true;module_param (log_ecn_error, bool,0644 );MODULE_PARM_DESC (log_ecn_error,"Log packets received with corrupted ECN" ); +static bool netns_atomic; +module_param (netns_atomic, bool,0444 ); +MODULE_PARM_DESC (netns_atomic, +"Create tunnel in target net namespace directly and use current net namespace as link-netns by default" ); +static struct rtnl_link_ops ipgre_link_ops__read_mostly ;static const struct header_ops ipgre_header_ops;
@@ -1393,6 +1398,7 @@ static int ipgre_newlink(struct net *src_net, struct net_device *dev, struct nlattr *tb[], struct nlattr *data[], struct netlink_ext_ack *extack) { + struct net *link_net = netns_atomic ? src_net : dev_net(dev);struct ip_tunnel_parm_kern p;__u32 fwmark =0 ;int err;
@@ -1404,13 +1410,14 @@ static int ipgre_newlink(struct net *src_net, struct net_device *dev, err =ipgre_netlink_parms (dev, data, tb, &p, &fwmark);if (err <0 )return err; -return ip_tunnel_newlink (dev, tb, &p, fwmark); +return ip_tunnel_newlink_net (link_net, dev, tb, &p, fwmark); }static int erspan_newlink (struct net *src_net,struct net_device *dev,struct nlattr *tb[],struct nlattr *data[],struct netlink_ext_ack *extack) { +struct net *link_net = netns_atomic ? src_net :dev_net (dev);struct ip_tunnel_parm_kern p;__u32 fwmark =0 ;int err;
@@ -1422,7 +1429,7 @@ static int erspan_newlink(struct net *src_net, struct net_device *dev, err =erspan_netlink_parms (dev, data, tb, &p, &fwmark);if (err)return err; -return ip_tunnel_newlink (dev, tb, &p, fwmark); +return ip_tunnel_newlink_net (link_net, dev, tb, &p, fwmark); } static int ipgre_changelink(struct net_device *dev, struct nlattr *tb[],
@@ -1777,6 +1784,10 @@ static int __init ipgre_init(void)pr_info ("GRE over IPv4 tunneling driver\n" ); + ipgre_link_ops.netns_atomic = netns_atomic; + ipgre_tap_ops.netns_atomic = netns_atomic; + erspan_link_ops.netns_atomic = netns_atomic; + err =register_pernet_device (&ipgre_net_ops);if (err <0 )return err;
Signature · 1 line
2.47.0
Jakub Kicinski
maintainerQuoted 10 lines · On Wed, 23 Oct 2024 10:31:41 +0800 Xiao Liang wrote:
This patch series includes some netns-related improvements and fixes for RTNL and
ip_tunnel, to make link creation more intuitive:- Creating link in another net namespace doesn't conflict with link names in current one.
- Add a flag inrtnl_ops, to avoid netns change when link-netns is present if possible.
- When creating ip tunnel (e.g. GRE) in another netns, use current as link-netns if not specified explicitly.
Quoted 4 lines
So that
# modprobe ip_gre netns_atomic=1 # ip link add netns ns1 link-netns ns2 tun0 type gre ...
Do you think the netns_atomic module param is really necessary? I doubt anyone cares about the event popping up in the wrong name space first.
BTW would be good to have tests for this. At least the behavior around name / ifindex collisions in different namespaces. You can possibly extend/re-purpose netns-name.sh for this.
For notifications you could use python and subscribe to the events using a YNL socket. May be easier than dealing with ip monitor as a background process. But either way is fine.
Quoted 12 lines · Jakub Kicinski
Quoted 10 lines · On Wed, 23 Oct 2024 10:31:41 +0800 Xiao Liang wrote:
This patch series includes some netns-related improvements and fixes for RTNL and
ip_tunnel, to make link creation more intuitive:- Creating link in another net namespace doesn't conflict with link names in current one.
- Add a flag inrtnl_ops, to avoid netns change when link-netns is present if possible.
- When creating ip tunnel (e.g. GRE) in another netns, use current as link-netns if not specified explicitly.
Quoted 8 lines
Quoted 4 lines
So that
# modprobe ip_gre netns_atomic=1 # ip link add netns ns1 link-netns ns2 tun0 type gre ...Do you think the
netns_atomicmodule param is really necessary? I doubt anyone cares about the event popping up in the wrong name space first.
We used FRRouting in our solution which listens to link notifications to set up corresponding objects in userspace. Since the events are sent in different namespaces (thus via different RTNL sockets), we can't guarantee that the events are received in the correct order, and have trouble processing them. The way to solve this problem I can think of is to have a multi-netns RTNL socket where all events are synchronized, or to eliminate the redundant events in the first place. The latter seems easier to implement.
Quoted 8 lines
BTW would be good to have tests for this. At least the behavior around name / ifindex collisions in different namespaces. You can possibly extend/re-purpose
netns-name.shfor this.For notifications you could use python and subscribe to the events using a YNL socket. May be easier than dealing with ip monitor as a background process. But either way is fine.
I will try to add some tests. Thanks!
Jakub Kicinski
maintainerQuoted 4 lines · On Wed, 30 Oct 2024 10:10:32 +0800 Xiao Liang wrote:
Quoted 3 lines
Do you think the
netns_atomicmodule param is really necessary? I doubt anyone cares about the event popping up in the wrong name space first.
Quoted 8 lines
We used FRRouting in our solution which listens to link notifications to set up corresponding objects in userspace. Since the events are sent in different namespaces (thus via different RTNL sockets), we can't guarantee that the events are received in the correct order, and have trouble processing them. The way to solve this problem I can think of is to have a multi-netns RTNL socket where all events are synchronized, or to eliminate the redundant events in the first place. The latter seems easier to implement.
I think we're on the same page. I'm saying that any potential user will run into a similar problem, and I don't see a clear use case for notifications in both namespaces. So we can try to make the behavior of netns_atomic=1 the default one and get rid of the module param.
Quoted 15 lines · Jakub Kicinski
Quoted 4 lines · On Wed, 30 Oct 2024 10:10:32 +0800 Xiao Liang wrote:
Quoted 3 lines
Do you think the
netns_atomicmodule param is really necessary? I doubt anyone cares about the event popping up in the wrong name space first.Quoted 8 lines
We used FRRouting in our solution which listens to link notifications to set up corresponding objects in userspace. Since the events are sent in different namespaces (thus via different RTNL sockets), we can't guarantee that the events are received in the correct order, and have trouble processing them. The way to solve this problem I can think of is to have a multi-netns RTNL socket where all events are synchronized, or to eliminate the redundant events in the first place. The latter seems easier to implement.
Quoted 5 lines
I think we're on the same page. I'm saying that any potential user will run into a similar problem, and I don't see a clear use case for notifications in both namespaces. So we can try to make the behavior of netns_atomic=1 the default one and get rid of the module param.
Ah.. I misunderstood your question... Besides notifications there's another behavior change.
In this patchset, RTNL core passes link-netns as src_net to drivers. But ip tunnel driver doesn't support source netns currently, and that's why we have Patch 4. As a side effect, source netns will be used if link-netns is not specified.For example:
ip -n ns1 link add netns ns2 gre1 type gre ...
In current implementation, the link-netns is ns2. While with netns_atomic=1, link-netns will be ns1 (source netns). The module param serves as a way to keep the current behavior.
I think we can make it default for drivers that already have source netns support.