diff mbox series

net: Allow to specify ifindex when device is moved to another namespace

Message ID 20210405071223.138101-1-avagin@gmail.com (mailing list archive)
State Superseded
Commit eeb85a14ee3494febb85ccfbee0772eda0823b13
Delegated to: Netdev Maintainers
Headers show
Series net: Allow to specify ifindex when device is moved to another namespace | expand

Checks

Context Check Description
netdev/cover_letter success Link
netdev/fixes_present success Link
netdev/patch_count success Link
netdev/tree_selection success Guessed tree name to be net-next
netdev/subject_prefix warning Target tree name not specified in the subject
netdev/cc_maintainers warning 24 maintainers not CCed: zhangkaiheb@126.com laniel_francis@privacyrequired.com johannes@sipsolutions.net linux-wpan@vger.kernel.org haiyangz@microsoft.com andriin@fb.com sthemmin@microsoft.com alex.aring@gmail.com linux-wireless@vger.kernel.org nikolay@nvidia.com atenart@kernel.org roopa@cumulusnetworks.com wei.liu@kernel.org zhudi21@huawei.com kys@microsoft.com andrew@lunn.ch daniel@iogearbox.net weiwan@google.com cong.wang@bytedance.com ast@kernel.org ap420073@gmail.com linux-hyperv@vger.kernel.org stefan@datenfreihafen.org edumazet@google.com
netdev/source_inline success Was 0 now: 0
netdev/verify_signedoff success Link
netdev/module_param success Was 0 now: 0
netdev/build_32bit success Errors and warnings before: 6680 this patch: 6680
netdev/kdoc success Errors and warnings before: 0 this patch: 0
netdev/verify_fixes success Link
netdev/checkpatch warning WARNING: line length of 83 exceeds 80 columns
netdev/build_allmodconfig_warn success Errors and warnings before: 6893 this patch: 6893
netdev/header_inline success Link

Commit Message

Andrei Vagin April 5, 2021, 7:12 a.m. UTC
Currently, we can specify ifindex on link creation. This change allows
to specify ifindex when a device is moved to another network namespace.

Even now, a device ifindex can be changed if there is another device
with the same ifindex in the target namespace. So this change doesn't
introduce completely new behavior, it adds more control to the process.

CRIU users want to restore containers with pre-created network devices.
A user will provide network devices and instructions where they have to
be restored, then CRIU will restore network namespaces and move devices
into them. The problem is that devices have to be restored with the same
indexes that they have before C/R.

Cc: Alexander Mikhalitsyn <alexander.mikhalitsyn@virtuozzo.com>
Suggested-by: Christian Brauner <christian.brauner@ubuntu.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
---
 drivers/net/hyperv/netvsc_drv.c |  2 +-
 include/linux/netdevice.h       |  3 ++-
 net/core/dev.c                  | 24 +++++++++++++++++-------
 net/core/rtnetlink.c            | 19 +++++++++++++++----
 net/ieee802154/core.c           |  4 ++--
 net/wireless/core.c             |  4 ++--
 6 files changed, 39 insertions(+), 17 deletions(-)

Comments

Christian Brauner April 5, 2021, 12:15 p.m. UTC | #1
On Mon, Apr 05, 2021 at 12:12:23AM -0700, Andrei Vagin wrote:
> Currently, we can specify ifindex on link creation. This change allows
> to specify ifindex when a device is moved to another network namespace.
> 
> Even now, a device ifindex can be changed if there is another device
> with the same ifindex in the target namespace. So this change doesn't
> introduce completely new behavior, it adds more control to the process.
> 
> CRIU users want to restore containers with pre-created network devices.
> A user will provide network devices and instructions where they have to
> be restored, then CRIU will restore network namespaces and move devices
> into them. The problem is that devices have to be restored with the same
> indexes that they have before C/R.
> 
> Cc: Alexander Mikhalitsyn <alexander.mikhalitsyn@virtuozzo.com>
> Suggested-by: Christian Brauner <christian.brauner@ubuntu.com>
> Signed-off-by: Andrei Vagin <avagin@gmail.com>
> ---

LGTM.
Thank for doing this!
Reviewed-by: Christian Brauner <christian.brauner@ubuntu.com>
patchwork-bot+netdevbpf@kernel.org April 5, 2021, 10:10 p.m. UTC | #2
Hello:

This patch was applied to netdev/net-next.git (refs/heads/master):

On Mon,  5 Apr 2021 00:12:23 -0700 you wrote:
> Currently, we can specify ifindex on link creation. This change allows
> to specify ifindex when a device is moved to another network namespace.
> 
> Even now, a device ifindex can be changed if there is another device
> with the same ifindex in the target namespace. So this change doesn't
> introduce completely new behavior, it adds more control to the process.
> 
> [...]

Here is the summary with links:
  - net: Allow to specify ifindex when device is moved to another namespace
    https://git.kernel.org/netdev/net-next/c/eeb85a14ee34

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
Jakub Kicinski April 5, 2021, 10:15 p.m. UTC | #3
On Mon,  5 Apr 2021 00:12:23 -0700 Andrei Vagin wrote:
> Currently, we can specify ifindex on link creation. This change allows
> to specify ifindex when a device is moved to another network namespace.
> 
> Even now, a device ifindex can be changed if there is another device
> with the same ifindex in the target namespace. So this change doesn't
> introduce completely new behavior, it adds more control to the process.
> 
> CRIU users want to restore containers with pre-created network devices.
> A user will provide network devices and instructions where they have to
> be restored, then CRIU will restore network namespaces and move devices
> into them. The problem is that devices have to be restored with the same
> indexes that they have before C/R.
> 
> Cc: Alexander Mikhalitsyn <alexander.mikhalitsyn@virtuozzo.com>
> Suggested-by: Christian Brauner <christian.brauner@ubuntu.com>
> Signed-off-by: Andrei Vagin <avagin@gmail.com>

> @@ -2354,7 +2354,7 @@ static int netvsc_register_vf(struct net_device *vf_netdev)
>  	 */
>  	if (!net_eq(dev_net(ndev), dev_net(vf_netdev))) {
>  		ret = dev_change_net_namespace(vf_netdev,
> -					       dev_net(ndev), "eth%d");
> +					       dev_net(ndev), "eth%d", 0);

Given vast majority of callers pass 0 as the new param - perhaps
dev_change_net_namespace() should become a static inline wrapper
over a function with more parameters?

>  		if (ret)
>  			netdev_err(vf_netdev,
>  				   "could not move to same namespace as %s: %d\n",

> diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
> index 1bdcb33fb561..d51252afde0a 100644
> --- a/net/core/rtnetlink.c
> +++ b/net/core/rtnetlink.c
> @@ -2266,6 +2266,9 @@ static int validate_linkmsg(struct net_device *dev, struct nlattr *tb[])
>  			return -EINVAL;
>  	}
>  
> +	if (tb[IFLA_NEW_IFINDEX] && nla_get_s32(tb[IFLA_NEW_IFINDEX]) <= 0)
> +		return -EINVAL;

I think you need to add IFLA_NEW_IFINDEX to ifla_policy, it used to be
an output only attribute, it's missing input validation. You can add
policy right there NLA_POLICY_MIN(NLA_S32, 0) - .min is 16 bit but it'd
get promoted correctly, I believe?

>  	if (tb[IFLA_AF_SPEC]) {
>  		struct nlattr *af;
>  		int rem, err;
diff mbox series

Patch

diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c
index 7349a70af083..8c0c70e1da77 100644
--- a/drivers/net/hyperv/netvsc_drv.c
+++ b/drivers/net/hyperv/netvsc_drv.c
@@ -2354,7 +2354,7 @@  static int netvsc_register_vf(struct net_device *vf_netdev)
 	 */
 	if (!net_eq(dev_net(ndev), dev_net(vf_netdev))) {
 		ret = dev_change_net_namespace(vf_netdev,
-					       dev_net(ndev), "eth%d");
+					       dev_net(ndev), "eth%d", 0);
 		if (ret)
 			netdev_err(vf_netdev,
 				   "could not move to same namespace as %s: %d\n",
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index f57b70fc251f..b482236c0e99 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -4026,7 +4026,8 @@  void __dev_notify_flags(struct net_device *, unsigned int old_flags,
 int dev_change_name(struct net_device *, const char *);
 int dev_set_alias(struct net_device *, const char *, size_t);
 int dev_get_alias(const struct net_device *, char *, size_t);
-int dev_change_net_namespace(struct net_device *, struct net *, const char *);
+int dev_change_net_namespace(struct net_device *dev, struct net *net,
+			     const char *pat, int new_ifindex);
 int __dev_set_mtu(struct net_device *, int);
 int dev_validate_mtu(struct net_device *dev, int mtu,
 		     struct netlink_ext_ack *extack);
diff --git a/net/core/dev.c b/net/core/dev.c
index b4c67a5be606..9d1a8fac793f 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -11067,6 +11067,8 @@  EXPORT_SYMBOL(unregister_netdev);
  *	@net: network namespace
  *	@pat: If not NULL name pattern to try if the current device name
  *	      is already taken in the destination network namespace.
+ *	@new_ifindex: If not zero, specifies device index in the target
+ *	              namespace.
  *
  *	This function shuts down a device interface and moves it
  *	to a new network namespace. On success 0 is returned, on
@@ -11075,10 +11077,11 @@  EXPORT_SYMBOL(unregister_netdev);
  *	Callers must hold the rtnl semaphore.
  */
 
-int dev_change_net_namespace(struct net_device *dev, struct net *net, const char *pat)
+int dev_change_net_namespace(struct net_device *dev, struct net *net,
+			     const char *pat, int new_ifindex)
 {
 	struct net *net_old = dev_net(dev);
-	int err, new_nsid, new_ifindex;
+	int err, new_nsid;
 
 	ASSERT_RTNL();
 
@@ -11109,6 +11112,11 @@  int dev_change_net_namespace(struct net_device *dev, struct net *net, const char
 			goto out;
 	}
 
+	/* Check that new_ifindex isn't used yet. */
+	err = -EBUSY;
+	if (new_ifindex && __dev_get_by_index(net, new_ifindex))
+		goto out;
+
 	/*
 	 * And now a mini version of register_netdevice unregister_netdevice.
 	 */
@@ -11136,10 +11144,12 @@  int dev_change_net_namespace(struct net_device *dev, struct net *net, const char
 
 	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);
-	else
-		new_ifindex = dev->ifindex;
+	if (!new_ifindex) {
+		if (__dev_get_by_index(net, dev->ifindex))
+			new_ifindex = dev_new_index(net);
+		else
+			new_ifindex = dev->ifindex;
+	}
 
 	rtmsg_ifinfo_newnet(RTM_DELLINK, dev, ~0U, GFP_KERNEL, &new_nsid,
 			    new_ifindex);
@@ -11448,7 +11458,7 @@  static void __net_exit default_device_exit(struct net *net)
 		snprintf(fb_name, IFNAMSIZ, "dev%d", dev->ifindex);
 		if (__dev_get_by_name(&init_net, fb_name))
 			snprintf(fb_name, IFNAMSIZ, "dev%%d");
-		err = dev_change_net_namespace(dev, &init_net, fb_name);
+		err = dev_change_net_namespace(dev, &init_net, fb_name, 0);
 		if (err) {
 			pr_emerg("%s: failed to move %s to init_net: %d\n",
 				 __func__, dev->name, err);
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index 1bdcb33fb561..d51252afde0a 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -2266,6 +2266,9 @@  static int validate_linkmsg(struct net_device *dev, struct nlattr *tb[])
 			return -EINVAL;
 	}
 
+	if (tb[IFLA_NEW_IFINDEX] && nla_get_s32(tb[IFLA_NEW_IFINDEX]) <= 0)
+		return -EINVAL;
+
 	if (tb[IFLA_AF_SPEC]) {
 		struct nlattr *af;
 		int rem, err;
@@ -2603,14 +2606,22 @@  static int do_setlink(const struct sk_buff *skb,
 		return err;
 
 	if (tb[IFLA_NET_NS_PID] || tb[IFLA_NET_NS_FD] || tb[IFLA_TARGET_NETNSID]) {
-		struct net *net = rtnl_link_get_net_capable(skb, dev_net(dev),
-							    tb, CAP_NET_ADMIN);
+		struct net *net;
+		int new_ifindex;
+
+		net = rtnl_link_get_net_capable(skb, dev_net(dev),
+						tb, CAP_NET_ADMIN);
 		if (IS_ERR(net)) {
 			err = PTR_ERR(net);
 			goto errout;
 		}
 
-		err = dev_change_net_namespace(dev, net, ifname);
+		if (tb[IFLA_NEW_IFINDEX])
+			new_ifindex = nla_get_s32(tb[IFLA_NEW_IFINDEX]);
+		else
+			new_ifindex = 0;
+
+		err = dev_change_net_namespace(dev, net, ifname, new_ifindex);
 		put_net(net);
 		if (err)
 			goto errout;
@@ -3452,7 +3463,7 @@  static int __rtnl_newlink(struct sk_buff *skb, struct nlmsghdr *nlh,
 	if (err < 0)
 		goto out_unregister;
 	if (link_net) {
-		err = dev_change_net_namespace(dev, dest_net, ifname);
+		err = dev_change_net_namespace(dev, dest_net, ifname, 0);
 		if (err < 0)
 			goto out_unregister;
 	}
diff --git a/net/ieee802154/core.c b/net/ieee802154/core.c
index de259b5170ab..ec3068937fc3 100644
--- a/net/ieee802154/core.c
+++ b/net/ieee802154/core.c
@@ -205,7 +205,7 @@  int cfg802154_switch_netns(struct cfg802154_registered_device *rdev,
 		if (!wpan_dev->netdev)
 			continue;
 		wpan_dev->netdev->features &= ~NETIF_F_NETNS_LOCAL;
-		err = dev_change_net_namespace(wpan_dev->netdev, net, "wpan%d");
+		err = dev_change_net_namespace(wpan_dev->netdev, net, "wpan%d", 0);
 		if (err)
 			break;
 		wpan_dev->netdev->features |= NETIF_F_NETNS_LOCAL;
@@ -222,7 +222,7 @@  int cfg802154_switch_netns(struct cfg802154_registered_device *rdev,
 				continue;
 			wpan_dev->netdev->features &= ~NETIF_F_NETNS_LOCAL;
 			err = dev_change_net_namespace(wpan_dev->netdev, net,
-						       "wpan%d");
+						       "wpan%d", 0);
 			WARN_ON(err);
 			wpan_dev->netdev->features |= NETIF_F_NETNS_LOCAL;
 		}
diff --git a/net/wireless/core.c b/net/wireless/core.c
index a2785379df6e..fabb677b7d58 100644
--- a/net/wireless/core.c
+++ b/net/wireless/core.c
@@ -165,7 +165,7 @@  int cfg80211_switch_netns(struct cfg80211_registered_device *rdev,
 		if (!wdev->netdev)
 			continue;
 		wdev->netdev->features &= ~NETIF_F_NETNS_LOCAL;
-		err = dev_change_net_namespace(wdev->netdev, net, "wlan%d");
+		err = dev_change_net_namespace(wdev->netdev, net, "wlan%d", 0);
 		if (err)
 			break;
 		wdev->netdev->features |= NETIF_F_NETNS_LOCAL;
@@ -182,7 +182,7 @@  int cfg80211_switch_netns(struct cfg80211_registered_device *rdev,
 				continue;
 			wdev->netdev->features &= ~NETIF_F_NETNS_LOCAL;
 			err = dev_change_net_namespace(wdev->netdev, net,
-							"wlan%d");
+							"wlan%d", 0);
 			WARN_ON(err);
 			wdev->netdev->features |= NETIF_F_NETNS_LOCAL;
 		}