[patch net-next-2.6] net: allow notifier subscribers to forbid device from closing

[patch net-next-2.6] net: allow notifier subscribers to forbid device from closing

[Jiri Pirko]

In some situations, like when the device is used as slave device in
bond/br/etc it is not nice if someone closes the device. This allows
it's masters to forbid this closure.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
---
 include/linux/netdevice.h |    1 +
 net/core/dev.c            |   34 ++++++++++++++++++++++++++++------
 net/core/rtnetlink.c      |    1 +
 3 files changed, 30 insertions(+), 6 deletions(-)

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index dad7e4d..b8047d3 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -1559,6 +1559,7 @@ struct packet_type {
 #define NETDEV_RELEASE		0x0012
 #define NETDEV_NOTIFY_PEERS	0x0013
 #define NETDEV_JOIN		0x0014
+#define NETDEV_PRE_DOWN		0x0015
 
 extern int register_netdevice_notifier(struct notifier_block *nb);
 extern int unregister_netdevice_notifier(struct notifier_block *nb);
diff --git a/net/core/dev.c b/net/core/dev.c
index 11b0fc7..d252a7e 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1253,11 +1253,24 @@ static int __dev_close_many(struct list_head *head)
 	return 0;
 }
 
+static int __dev_pre_close(struct net_device *dev)
+{
+	int err;
+
+	err = call_netdevice_notifiers(NETDEV_PRE_DOWN, dev);
+	if (err)
+		return notifier_to_errno(err);
+	return 0;
+}
+
 static int __dev_close(struct net_device *dev)
 {
 	int retval;
 	LIST_HEAD(single);
 
+	retval = __dev_pre_close(dev);
+	if (retval)
+		return retval;
 	list_add(&dev->unreg_list, &single);
 	retval = __dev_close_many(&single);
 	list_del(&single);
@@ -1269,9 +1282,12 @@ static int dev_close_many(struct list_head *head)
 	struct net_device *dev, *tmp;
 	LIST_HEAD(tmp_list);
 
-	list_for_each_entry_safe(dev, tmp, head, unreg_list)
+	list_for_each_entry_safe(dev, tmp, head, unreg_list) {
 		if (!(dev->flags & IFF_UP))
 			list_move(&dev->unreg_list, &tmp_list);
+		else
+			__dev_pre_close(dev);
+	}
 
 	__dev_close_many(head);
 
@@ -1289,21 +1305,26 @@ static int dev_close_many(struct list_head *head)
  *	dev_close - shutdown an interface.
  *	@dev: device to shutdown
  *
- *	This function moves an active device into down state. A
- *	%NETDEV_GOING_DOWN is sent to the netdev notifier chain. The device
- *	is then deactivated and finally a %NETDEV_DOWN is sent to the notifier
- *	chain.
+ *	This function moves an active device into down state.
+ *	A %NETDEV_PRE_DOWN and %NETDEV_GOING_DOWN is sent to the netdev
+ *	notifier chain. The device is then deactivated and finally
+ *	a %NETDEV_DOWN is sent to the notifier chain.
  */
 int dev_close(struct net_device *dev)
 {
+	int retval = 0;
+
 	if (dev->flags & IFF_UP) {
 		LIST_HEAD(single);
 
+		retval = __dev_pre_close(dev);
+		if (retval)
+			return retval;
 		list_add(&dev->unreg_list, &single);
 		dev_close_many(&single);
 		list_del(&single);
 	}
-	return 0;
+	return retval;
 }
 EXPORT_SYMBOL(dev_close);
 
@@ -1397,6 +1418,7 @@ rollback:
 				break;
 
 			if (dev->flags & IFF_UP) {
+				nb->notifier_call(nb, NETDEV_PRE_DOWN, dev);
 				nb->notifier_call(nb, NETDEV_GOING_DOWN, dev);
 				nb->notifier_call(nb, NETDEV_DOWN, dev);
 			}
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index 39f8dd6..34f5b32 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -1997,6 +1997,7 @@ static int rtnetlink_event(struct notifier_block *this, unsigned long event, voi
 	case NETDEV_UP:
 	case NETDEV_DOWN:
 	case NETDEV_PRE_UP:
+	case NETDEV_PRE_DOWN:
 	case NETDEV_POST_INIT:
 	case NETDEV_REGISTER:
 	case NETDEV_CHANGE:
-- 
1.7.6

Re: [patch net-next-2.6] net: allow notifier subscribers to forbid device from closing

[Stephen Hemminger]

On Wed, 31 Aug 2011 17:15:31 +0200
Jiri Pirko <jpirko@redhat.com> wrote:

> In some situations, like when the device is used as slave device in
> bond/br/etc it is not nice if someone closes the device. This allows
> it's masters to forbid this closure.
> 
> Signed-off-by: Jiri Pirko <jpirko@redhat.com>

I don't think this is necessary, for bridging case.
bridging handles it. And bonding should as well.
It is a good way to test STP etc.

Is this a case of "you really shouldn't do this", or
"don't do this it will crash"? In general Linux allows the
former and uses references to prevent the later.

Re: [patch net-next-2.6] net: allow notifier subscribers to forbid device from closing

[Ben Hutchings]

On Wed, 2011-08-31 at 17:15 +0200, Jiri Pirko wrote:
> In some situations, like when the device is used as slave device in
> bond/br/etc it is not nice if someone closes the device. This allows
> it's masters to forbid this closure.

No it doesn't.

[...]
> @@ -1269,9 +1282,12 @@ static int dev_close_many(struct list_head *head)
>  	struct net_device *dev, *tmp;
>  	LIST_HEAD(tmp_list);
>  
> -	list_for_each_entry_safe(dev, tmp, head, unreg_list)
> +	list_for_each_entry_safe(dev, tmp, head, unreg_list) {
>  		if (!(dev->flags & IFF_UP))
>  			list_move(&dev->unreg_list, &tmp_list);
> +		else
> +			__dev_pre_close(dev);
> +	}
>  
>  	__dev_close_many(head);

The return value is ignored here.

And this is called from dev_close(), where you are adding the
notification as well.  So the notifier will usually be called twice.

[...]
> @@ -1397,6 +1418,7 @@ rollback:
>  				break;
>  
>  			if (dev->flags & IFF_UP) {
> +				nb->notifier_call(nb, NETDEV_PRE_DOWN, dev);
>  				nb->notifier_call(nb, NETDEV_GOING_DOWN, dev);
>  				nb->notifier_call(nb, NETDEV_DOWN, dev);
>  			}
[...]

The return value has to be ignored here.  Not sure it makes any sense to
call the notifier at all.

Ben.

-- 
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.

Re: [patch net-next-2.6] net: allow notifier subscribers to forbid device from closing

[Stephen Hemminger]

On Wed, 31 Aug 2011 16:25:51 +0100
Ben Hutchings <bhutchings@solarflare.com> wrote:

> On Wed, 2011-08-31 at 17:15 +0200, Jiri Pirko wrote:
> > In some situations, like when the device is used as slave device in
> > bond/br/etc it is not nice if someone closes the device. This allows
> > it's masters to forbid this closure.
> 
> No it doesn't.
> 
> [...]
> > @@ -1269,9 +1282,12 @@ static int dev_close_many(struct list_head *head)
> >  	struct net_device *dev, *tmp;
> >  	LIST_HEAD(tmp_list);
> >  
> > -	list_for_each_entry_safe(dev, tmp, head, unreg_list)
> > +	list_for_each_entry_safe(dev, tmp, head, unreg_list) {
> >  		if (!(dev->flags & IFF_UP))
> >  			list_move(&dev->unreg_list, &tmp_list);
> > +		else
> > +			__dev_pre_close(dev);
> > +	}
> >  
> >  	__dev_close_many(head);
> 
> The return value is ignored here.
> 
> And this is called from dev_close(), where you are adding the
> notification as well.  So the notifier will usually be called twice.
> 
> [...]
> > @@ -1397,6 +1418,7 @@ rollback:
> >  				break;
> >  
> >  			if (dev->flags & IFF_UP) {
> > +				nb->notifier_call(nb, NETDEV_PRE_DOWN, dev);
> >  				nb->notifier_call(nb, NETDEV_GOING_DOWN, dev);
> >  				nb->notifier_call(nb, NETDEV_DOWN, dev);
> >  			}
> [...]
> 
> The return value has to be ignored here.  Not sure it makes any sense to
> call the notifier at all.
> 
> Ben.
> 

Also we need to allow rmmod'ing a network device even it is
part of a bridge and that implicitly
calls close.

Re: [patch net-next-2.6] net: allow notifier subscribers to forbid device from closing

[Jiri Pirko]

Wed, Aug 31, 2011 at 05:25:51PM CEST, bhutchings@solarflare.com wrote:
>On Wed, 2011-08-31 at 17:15 +0200, Jiri Pirko wrote:
>> In some situations, like when the device is used as slave device in
>> bond/br/etc it is not nice if someone closes the device. This allows
>> it's masters to forbid this closure.
>
>No it doesn't.
It does

>
>[...]
>> @@ -1269,9 +1282,12 @@ static int dev_close_many(struct list_head *head)
>>  	struct net_device *dev, *tmp;
>>  	LIST_HEAD(tmp_list);
>>  
>> -	list_for_each_entry_safe(dev, tmp, head, unreg_list)
>> +	list_for_each_entry_safe(dev, tmp, head, unreg_list) {
>>  		if (!(dev->flags & IFF_UP))
>>  			list_move(&dev->unreg_list, &tmp_list);
>> +		else
>> +			__dev_pre_close(dev);
>> +	}
>>  
>>  	__dev_close_many(head);
>
>The return value is ignored here.

That's intended. The reason is this is called from
rollback_registered_many - refuse should be ignored in that case
>
>And this is called from dev_close(), where you are adding the
>notification as well.  So the notifier will usually be called twice.
>
Indeed. Anyway I thought about it and we probably do not need this patch
as Stephen said.

>[...]
>> @@ -1397,6 +1418,7 @@ rollback:
>>  				break;
>>  
>>  			if (dev->flags & IFF_UP) {
>> +				nb->notifier_call(nb, NETDEV_PRE_DOWN, dev);
>>  				nb->notifier_call(nb, NETDEV_GOING_DOWN, dev);
>>  				nb->notifier_call(nb, NETDEV_DOWN, dev);
>>  			}
>[...]
>
>The return value has to be ignored here.  Not sure it makes any sense to
>call the notifier at all.
>
>Ben.
>
>-- 
>Ben Hutchings, Staff Engineer, Solarflare
>Not speaking for my employer; that's the marketing department's job.
>They asked us to note that Solarflare product names are trademarked.
>

Re: [patch net-next-2.6] net: allow notifier subscribers to forbid device from closing

[Jiri Pirko]

Wed, Aug 31, 2011 at 05:53:38PM CEST, shemminger@vyatta.com wrote:
>On Wed, 31 Aug 2011 16:25:51 +0100
>Ben Hutchings <bhutchings@solarflare.com> wrote:
>
>> On Wed, 2011-08-31 at 17:15 +0200, Jiri Pirko wrote:
>> > In some situations, like when the device is used as slave device in
>> > bond/br/etc it is not nice if someone closes the device. This allows
>> > it's masters to forbid this closure.
>> 
>> No it doesn't.
>> 
>> [...]
>> > @@ -1269,9 +1282,12 @@ static int dev_close_many(struct list_head *head)
>> >  	struct net_device *dev, *tmp;
>> >  	LIST_HEAD(tmp_list);
>> >  
>> > -	list_for_each_entry_safe(dev, tmp, head, unreg_list)
>> > +	list_for_each_entry_safe(dev, tmp, head, unreg_list) {
>> >  		if (!(dev->flags & IFF_UP))
>> >  			list_move(&dev->unreg_list, &tmp_list);
>> > +		else
>> > +			__dev_pre_close(dev);
>> > +	}
>> >  
>> >  	__dev_close_many(head);
>> 
>> The return value is ignored here.
>> 
>> And this is called from dev_close(), where you are adding the
>> notification as well.  So the notifier will usually be called twice.
>> 
>> [...]
>> > @@ -1397,6 +1418,7 @@ rollback:
>> >  				break;
>> >  
>> >  			if (dev->flags & IFF_UP) {
>> > +				nb->notifier_call(nb, NETDEV_PRE_DOWN, dev);
>> >  				nb->notifier_call(nb, NETDEV_GOING_DOWN, dev);
>> >  				nb->notifier_call(nb, NETDEV_DOWN, dev);
>> >  			}
>> [...]
>> 
>> The return value has to be ignored here.  Not sure it makes any sense to
>> call the notifier at all.
>> 
>> Ben.
>> 
>
>Also we need to allow rmmod'ing a network device even it is
>part of a bridge and that implicitly
>calls close.


this is not a problem because when it's called from
rollback_registered_many, return value if pre_down is ignored.