Virtualization - Apr 2019 - [PATCH net v6] failover: allow name change on IFF_UP slave interfaces

On Wed, Apr 03, 2019 at 12:52:47AM -0400, Si-Wei Liu
wrote:
> When a netdev appears through hot plug then gets enslaved by a failover
> master that is already up and running, the slave will be opened
> right away after getting enslaved. Today there's a race that userspace
> (udev) may fail to rename the slave if the kernel (net_failover)
> opens the slave earlier than when the userspace rename happens.
> Unlike bond or team, the primary slave of failover can't be renamed by
> userspace ahead of time, since the kernel initiated auto-enslavement is
> unable to, or rather, is never meant to be synchronized with the rename
> request from userspace.
> 
> As the failover slave interfaces are not designed to be operated
> directly by userspace apps: IP configuration, filter rules with
> regard to network traffic passing and etc., should all be done on master
> interface. In general, userspace apps only care about the
> name of master interface, while slave names are less important as long
> as admin users can see reliable names that may carry
> other information describing the netdev. For e.g., they can infer that
> "ens3nsby" is a standby slave of "ens3", while for a
> name like "eth0" they can't tell which master it belongs to.
> 
> Historically the name of IFF_UP interface can't be changed because
> there might be admin script or management software that is already
> relying on such behavior and assumes that the slave name can't be
> changed once UP. But failover is special: with the in-kernel
> auto-enslavement mechanism, the userspace expectation for device
> enumeration and bring-up order is already broken. Previously initramfs
> and various userspace config tools were modified to bypass failover
> slaves because of auto-enslavement and duplicate MAC address. Similarly,
> in case that users care about seeing reliable slave name, the new type
> of failover slaves needs to be taken care of specifically in userspace
> anyway.
> 
> It's less risky to lift up the rename restriction on failover slave
> which is already UP. Although it's possible this change may potentially
> break userspace component (most likely configuration scripts or
> management software) that assumes slave name can't be changed while
> UP, it's relatively a limited and controllable set among all userspace
> components, which can be fixed specifically to listen for the rename
> and/or link down/up events on failover slaves. Userspace component
> interacting with slaves is expected to be changed to operate on failover
> master interface instead, as the failover slave is dynamic in nature
> which may come and go at any point.  The goal is to make the role of
> failover slaves less relevant, and userspace components should only
> deal with failover master in the long run.
> 
> Fixes: 30c8bd5aa8b2 ("net: Introduce generic failover module")
> Signed-off-by: Si-Wei Liu <si-wei.liu at oracle.com>
> Reviewed-by: Liran Alon <liran.alon at oracle.com>
Acked-by: Michael S. Tsirkin <mst at redhat.com>

Stephen are you happy with this approach?
> --
> v1 -> v2:
> - Drop configurable module parameter (Sridhar)
> 
> v2 -> v3:
> - Drop additional IFF_SLAVE_RENAME_OK flag (Sridhar)
> - Send down and up events around rename (Michael S. Tsirkin)
> 
> v3 -> v4:
> - Simplify notification to be sent (Stephen Hemminger)
> 
> v4 -> v5:
> - Sync up code with latest net-next (Sridhar)
> - Use proper structure initialization (Stephen, Jiri)
> 
> v5 -> v6:
> - Make the property of live name change a generic flag (Stephen)
> ---
>  include/linux/netdevice.h |  3 +++
>  net/core/dev.c            | 25 ++++++++++++++++++++++++-
>  net/core/failover.c       |  6 +++---
>  3 files changed, 30 insertions(+), 4 deletions(-)
> 
> diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
> index 78f5ec4e..ea9a63f 100644
> --- a/include/linux/netdevice.h
> +++ b/include/linux/netdevice.h
> @@ -1498,6 +1498,7 @@ struct net_device_ops {
>   * @IFF_FAILOVER: device is a failover master device
>   * @IFF_FAILOVER_SLAVE: device is lower dev of a failover master device
>   * @IFF_L3MDEV_RX_HANDLER: only invoke the rx handler of L3 master device
> + * @IFF_LIVE_NAME_CHANGE: rename is allowed while device is running
>   */
>  enum netdev_priv_flags {
>  	IFF_802_1Q_VLAN			= 1<<0,
> @@ -1530,6 +1531,7 @@ enum netdev_priv_flags {
>  	IFF_FAILOVER			= 1<<27,
>  	IFF_FAILOVER_SLAVE		= 1<<28,
>  	IFF_L3MDEV_RX_HANDLER		= 1<<29,
> +	IFF_LIVE_NAME_CHANGE		= 1<<30,
>  };
>  
>  #define IFF_802_1Q_VLAN			IFF_802_1Q_VLAN
> @@ -1561,6 +1563,7 @@ enum netdev_priv_flags {
>  #define IFF_FAILOVER			IFF_FAILOVER
>  #define IFF_FAILOVER_SLAVE		IFF_FAILOVER_SLAVE
>  #define IFF_L3MDEV_RX_HANDLER		IFF_L3MDEV_RX_HANDLER
> +#define IFF_LIVE_NAME_CHANGE		IFF_LIVE_NAME_CHANGE
>  
>  /**
>   *	struct net_device - The DEVICE structure.
> diff --git a/net/core/dev.c b/net/core/dev.c
> index 9823b77..48341d5 100644
> --- a/net/core/dev.c
> +++ b/net/core/dev.c
> @@ -1185,7 +1185,21 @@ int dev_change_name(struct net_device *dev, const
char *newname)
>  	BUG_ON(!dev_net(dev));
>  
>  	net = dev_net(dev);
> -	if (dev->flags & IFF_UP)
> +
> +	/* Some auto-enslaved devices e.g. failover slaves are
> +	 * special, as userspace might rename the device after
> +	 * the interface had been brought up and running since
> +	 * the point kernel initiated auto-enslavement. Allow
> +	 * live name change even when these slave devices are
> +	 * up and running.
> +	 *
> +	 * Typically, users of these auto-enslaving devices
> +	 * don't actually care about slave name change, as
> +	 * they are supposed to operate on master interface
> +	 * directly.
> +	 */
> +	if (dev->flags & IFF_UP &&
> +	    likely(!(dev->priv_flags & IFF_LIVE_NAME_CHANGE)))
>  		return -EBUSY;
>  
>  	write_seqcount_begin(&devnet_rename_seq);
> @@ -1232,6 +1246,15 @@ int dev_change_name(struct net_device *dev, const
char *newname)
>  	hlist_add_head_rcu(&dev->name_hlist, dev_name_hash(net,
dev->name));
>  	write_unlock_bh(&dev_base_lock);
>  
> +	if (unlikely(dev->flags & IFF_UP)) {
> +		struct netdev_notifier_change_info change_info = {
> +			.info.dev = dev,
> +		};
> +
> +		call_netdevice_notifiers_info(NETDEV_CHANGE,
> +					      &change_info.info);
> +	}
> +
>  	ret = call_netdevice_notifiers(NETDEV_CHANGENAME, dev);
>  	ret = notifier_to_errno(ret);
>  
> diff --git a/net/core/failover.c b/net/core/failover.c
> index 4a92a98..b5cd3c7 100644
> --- a/net/core/failover.c
> +++ b/net/core/failover.c
> @@ -80,14 +80,14 @@ static int failover_slave_register(struct net_device
*slave_dev)
>  		goto err_upper_link;
>  	}
>  
> -	slave_dev->priv_flags |= IFF_FAILOVER_SLAVE;
> +	slave_dev->priv_flags |= (IFF_FAILOVER_SLAVE | IFF_LIVE_NAME_CHANGE);
>  
>  	if (fops && fops->slave_register &&
>  	    !fops->slave_register(slave_dev, failover_dev))
>  		return NOTIFY_OK;
>  
>  	netdev_upper_dev_unlink(slave_dev, failover_dev);
> -	slave_dev->priv_flags &= ~IFF_FAILOVER_SLAVE;
> +	slave_dev->priv_flags &= ~(IFF_FAILOVER_SLAVE |
IFF_LIVE_NAME_CHANGE);
>  err_upper_link:
>  	netdev_rx_handler_unregister(slave_dev);
>  done:
> @@ -121,7 +121,7 @@ int failover_slave_unregister(struct net_device
*slave_dev)
>  
>  	netdev_rx_handler_unregister(slave_dev);
>  	netdev_upper_dev_unlink(slave_dev, failover_dev);
> -	slave_dev->priv_flags &= ~IFF_FAILOVER_SLAVE;
> +	slave_dev->priv_flags &= ~(IFF_FAILOVER_SLAVE |
IFF_LIVE_NAME_CHANGE);
>  
>  	if (fops && fops->slave_unregister &&
>  	    !fops->slave_unregister(slave_dev, failover_dev))
> -- 
> 1.8.3.1

On Fri, 5 Apr 2019 17:28:55 -0400
"Michael S. Tsirkin" <mst at redhat.com> wrote:
> On Wed, Apr 03, 2019 at 12:52:47AM -0400, Si-Wei Liu wrote:
> > When a netdev appears through hot plug then gets enslaved by a
failover
> > master that is already up and running, the slave will be opened
> > right away after getting enslaved. Today there's a race that
userspace
> > (udev) may fail to rename the slave if the kernel (net_failover)
> > opens the slave earlier than when the userspace rename happens.
> > Unlike bond or team, the primary slave of failover can't be
renamed by
> > userspace ahead of time, since the kernel initiated auto-enslavement
is
> > unable to, or rather, is never meant to be synchronized with the
rename
> > request from userspace.
> > 
> > As the failover slave interfaces are not designed to be operated
> > directly by userspace apps: IP configuration, filter rules with
> > regard to network traffic passing and etc., should all be done on
master
> > interface. In general, userspace apps only care about the
> > name of master interface, while slave names are less important as long
> > as admin users can see reliable names that may carry
> > other information describing the netdev. For e.g., they can infer that
> > "ens3nsby" is a standby slave of "ens3", while for
a
> > name like "eth0" they can't tell which master it belongs
to.
> > 
> > Historically the name of IFF_UP interface can't be changed because
> > there might be admin script or management software that is already
> > relying on such behavior and assumes that the slave name can't be
> > changed once UP. But failover is special: with the in-kernel
> > auto-enslavement mechanism, the userspace expectation for device
> > enumeration and bring-up order is already broken. Previously initramfs
> > and various userspace config tools were modified to bypass failover
> > slaves because of auto-enslavement and duplicate MAC address.
Similarly,
> > in case that users care about seeing reliable slave name, the new type
> > of failover slaves needs to be taken care of specifically in userspace
> > anyway.
> > 
> > It's less risky to lift up the rename restriction on failover
slave
> > which is already UP. Although it's possible this change may
potentially
> > break userspace component (most likely configuration scripts or
> > management software) that assumes slave name can't be changed
while
> > UP, it's relatively a limited and controllable set among all
userspace
> > components, which can be fixed specifically to listen for the rename
> > and/or link down/up events on failover slaves. Userspace component
> > interacting with slaves is expected to be changed to operate on
failover
> > master interface instead, as the failover slave is dynamic in nature
> > which may come and go at any point.  The goal is to make the role of
> > failover slaves less relevant, and userspace components should only
> > deal with failover master in the long run.
> > 
> > Fixes: 30c8bd5aa8b2 ("net: Introduce generic failover
module")
> > Signed-off-by: Si-Wei Liu <si-wei.liu at oracle.com>
> > Reviewed-by: Liran Alon <liran.alon at oracle.com>  
> 
> Acked-by: Michael S. Tsirkin <mst at redhat.com>
> 
> Stephen are you happy with this approach?
I think it is the best solution for what you want to do. 

Did you test with some things like Free Range Routing, VPP or other userspace
control planes that consume netlink?

On 4/6/2019 12:21 AM, si-wei liu wrote:
>>>
>>> Stephen are you happy with this approach?
>> I think it is the best solution for what you want to do.
> 
> Since you're asking specifically, I tried what you suggested below.
> 
>> Did you test with some things like Free Range Routing,
> 
> Although there might be spurious warning (which is a check for sanity 
> more than an error) while slave interface is up, slave rename had been 
> handled quite well there, no matter which state slave is at.
> 
> https://github.com/FRRouting/frr/blob/master/zebra/if_netlink.c#L97
> 
> The FRR users are supposed to operate on failover master interface 
> anyway. No one is expected to configure those passive interfaces for 
> routing.
> 
>> VPP
> Nothing particular was seen for this one. The netlink usage there 
> doesn't seem related to my change:
> https://github.com/FDio/vpp/blob/master/src/vnet/devices/netlink.c
> 
>> or other userspace
>> control planes that consume netlink?
> dhcpcd (https://github.com/kobolabs/dhcpcd/blob/kobo/if-linux.c#L761) 
> was tested OK.
> 
> In addition, the patch seems to play quite well with systemd-udev and 
> dracut/initramfs-tools. No breakage, no weird error message was seen.
> 
> What else do you suggest we should try/test with?
Thanks Siwei for all the tests you are trying out. Did you notice that 
any of these tests required the NETDEV_CHANGE notifier that you added?

-Sridhar