Re: [PATCH net-next v2] bridge: also trigger RTM_NEWLINK when interface is released from bridge

[Vincent Bernat]

 ❦ 21 septembre 2017 08:09 -0700, Roopa Prabhu  :

>>> The one concern is that ports added or removed through ioctl should
>>> cause same events as doing the same thing via netlink. Some users use
>>> brctl (ioctl) and others use newer bridge (netlink) API.
>>
>> I'll make a third iteration to have the same notifications when using
>> ioctl() with details in the commit message.
>> --
>
> as long as the ioctl path for bridge port removal is generating a:
> RTM_DELLINK with AF_BRIDGE and
> RTM_NEWLINK with AF_UNSPEC with 'master' removed
>
> we should be good. These are the most important ones.
>
> are there other specific bridge-events missing ?. you might want to
> run `bridge monitor link` also. and un-slaving of a port also triggers
> fdb events which are visible under `bridge monitor fdb`

With the patch, bridge monitor link generates the same event with
ioctl() than with netlink (like for ip monitor link, netlink generates
one extra duplicate event when enslaving).

For bridge monitor fdb, there is a difference. With ioctl(), I don't get
the event for VLAN1:

Deleted ca:18:06:bc:f6:11 dev dummy1 vlan 1 master bridge0 permanent

I suppose this is an expected difference due to the inability to manage
VLAN-aware bridges with ioctl().
-- 
Use the fundamental control flow constructs.
- The Elements of Programming Style (Kernighan & Plauger)

Re: [PATCH net-next v2] bridge: also trigger RTM_NEWLINK when interface is released from bridge

[Roopa Prabhu]

On Thu, Sep 21, 2017 at 3:04 AM, Vincent Bernat  wrote:
>  ❦ 20 septembre 2017 16:21 -0700, Stephen Hemminger 
>  :
>
>> The one concern is that ports added or removed through ioctl should
>> cause same events as doing the same thing via netlink. Some users use
>> brctl (ioctl) and others use newer bridge (netlink) API.
>
> I'll make a third iteration to have the same notifications when using
> ioctl() with details in the commit message.
> --

as long as the ioctl path for bridge port removal is generating a:
RTM_DELLINK with AF_BRIDGE and
RTM_NEWLINK with AF_UNSPEC with 'master' removed

we should be good. These are the most important ones.

are there other specific bridge-events missing ?. you might want to
run `bridge monitor link` also. and un-slaving of a port also triggers
fdb events which are visible under `bridge monitor fdb`

Re: [PATCH net-next v2] bridge: also trigger RTM_NEWLINK when interface is released from bridge

[Vincent Bernat]

 ❦ 20 septembre 2017 16:21 -0700, Stephen Hemminger 
 :

> The one concern is that ports added or removed through ioctl should
> cause same events as doing the same thing via netlink. Some users use
> brctl (ioctl) and others use newer bridge (netlink) API.

I'll make a third iteration to have the same notifications when using
ioctl() with details in the commit message.
-- 
When in doubt, tell the truth.
-- Mark Twain

Re: [PATCH net-next v2] bridge: also trigger RTM_NEWLINK when interface is released from bridge

[Stephen Hemminger]

On Wed, 20 Sep 2017 15:57:16 -0600
David Ahern  wrote:

> On 9/20/17 3:09 PM, David Miller wrote:
> > From: Vincent Bernat 
> > Date: Sat, 16 Sep 2017 16:18:33 +0200
> > 
> > David, I am CC:'ing you because you've done work in this area over the
> > past year.  I'm applying this patch, it's been sitting since the 16th
> > and likes entirely correct to me.  But if you have objections just let
> > me know.
> >   
> >> Currently, when an interface is released from a bridge via
> >> ioctl(), we get a RTM_DELLINK event through netlink:
> >>
> >> Deleted 2: dummy0:  mtu 1500 master bridge0 
> >> state UNKNOWN
> >> link/ether 6e:23:c2:54:3a:b3
> >>
> >> Userspace has to interpret that as a removal from the bridge, not as a
> >> complete removal of the interface. When an bridged interface is
> >> completely removed, we get two events:
> >>
> >> Deleted 2: dummy0:  mtu 1500 master bridge0 state DOWN
> >> link/ether 6e:23:c2:54:3a:b3
> >> Deleted 2: dummy0:  mtu 1500 qdisc noop state DOWN group 
> >> default
> >> link/ether 6e:23:c2:54:3a:b3 brd ff:ff:ff:ff:ff:ff
> >>
> >> In constrast, when an interface is released from a bond, we get a
> >> RTM_NEWLINK with only the new characteristics (no master):
> >>
> >> 3: dummy1:  mtu 1500 qdisc noqueue 
> >> master bond0 state UNKNOWN group default
> >> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
> >> 3: dummy1:  mtu 1500 qdisc noqueue state 
> >> UNKNOWN group default
> >> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
> >> 4: bond0:  mtu 1500 qdisc noqueue 
> >> state UP group default
> >> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
> >> 3: dummy1:  mtu 1500 qdisc noqueue state DOWN group 
> >> default
> >> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
> >> 3: dummy1:  mtu 1500 qdisc noqueue state DOWN group 
> >> default
> >> link/ether ca:c8:7b:66:f8:25 brd ff:ff:ff:ff:ff:ff
> >> 4: bond0:  mtu 1500 qdisc noqueue 
> >> state UP group default
> >> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
> >>
> >> Userland may be confused by the fact we say a link is deleted while
> >> its characteristics are only modified. A first solution would have
> >> been to turn the RTM_DELLINK event in del_nbp() into a RTM_NEWLINK
> >> event. However, maybe some piece of userland is relying on this
> >> RTM_DELLINK to detect when a bridged interface is released. Instead,
> >> we also emit a RTM_NEWLINK event once the interface is
> >> released (without master info).
> >>
> >> Deleted 2: dummy0:  mtu 1500 master bridge0 
> >> state UNKNOWN
> >> link/ether 8a:bb:e7:94:b1:f8
> >> 2: dummy0:  mtu 1500 qdisc noqueue state 
> >> UNKNOWN group default
> >> link/ether 8a:bb:e7:94:b1:f8 brd ff:ff:ff:ff:ff:ff
> >>
> >> This is done only when using ioctl(). When using Netlink, such an
> >> event is already automatically emitted in do_setlink().  
> 
> The DELLINK is for AF_BRIDGE family (ifi_family). Adding family to
> print_linkinfo and running the monitor I get:
> 
> 
> [LINK]family 0: 35: dummy1:  mtu 1500 qdisc
> noqueue master br0 state UNKNOWN group default
> link/ether d6:c3:73:86:3c:73 brd ff:ff:ff:ff:ff:ff
> 
> [LINK]family 7: 35: dummy1:  mtu 1500
> master br0 state UNKNOWN
> link/ether d6:c3:73:86:3c:73
> 
> [LINK]Deleted family 7: 35: dummy1:  mtu
> 1500 master br0 state UNKNOWN
> link/ether d6:c3:73:86:3c:73
> 
> [LINK]family 0: 35: dummy1:  mtu 1500 qdisc
> noqueue state UNKNOWN group default
> link/ether d6:c3:73:86:3c:73 brd ff:ff:ff:ff:ff:ff
> 
> And that seems correct. So I think the RTM_NEWLINK added by this patch
> should not be needed.

Agreed, thanks for tracing this.

The one concern is that ports added or removed through ioctl should
cause same events as doing the same thing via netlink. Some users use
brctl (ioctl) and others use newer bridge (netlink) API.

Re: [PATCH net-next v2] bridge: also trigger RTM_NEWLINK when interface is released from bridge

[David Miller]

From: Vincent Bernat 
Date: Thu, 21 Sep 2017 00:12:53 +0200

>  ❦ 20 septembre 2017 15:57 -0600, David Ahern  :
> 
>> The DELLINK is for AF_BRIDGE family (ifi_family). Adding family to
>> print_linkinfo and running the monitor I get:
>>
>>
>> [LINK]family 0: 35: dummy1:  mtu 1500 qdisc
>> noqueue master br0 state UNKNOWN group default
>> link/ether d6:c3:73:86:3c:73 brd ff:ff:ff:ff:ff:ff
>>
>> [LINK]family 7: 35: dummy1:  mtu 1500
>> master br0 state UNKNOWN
>> link/ether d6:c3:73:86:3c:73
>>
>> [LINK]Deleted family 7: 35: dummy1:  mtu
>> 1500 master br0 state UNKNOWN
>> link/ether d6:c3:73:86:3c:73
> 
> I didn't know about the family. We can drop the patch.

Ok, I've reverted.

Thanks.

Re: [PATCH net-next v2] bridge: also trigger RTM_NEWLINK when interface is released from bridge

[Vincent Bernat]

 ❦ 20 septembre 2017 15:57 -0600, David Ahern  :

> The DELLINK is for AF_BRIDGE family (ifi_family). Adding family to
> print_linkinfo and running the monitor I get:
>
>
> [LINK]family 0: 35: dummy1:  mtu 1500 qdisc
> noqueue master br0 state UNKNOWN group default
> link/ether d6:c3:73:86:3c:73 brd ff:ff:ff:ff:ff:ff
>
> [LINK]family 7: 35: dummy1:  mtu 1500
> master br0 state UNKNOWN
> link/ether d6:c3:73:86:3c:73
>
> [LINK]Deleted family 7: 35: dummy1:  mtu
> 1500 master br0 state UNKNOWN
> link/ether d6:c3:73:86:3c:73

I didn't know about the family. We can drop the patch.
-- 
One of the most striking differences between a cat and a lie is that a cat has
only nine lives.
-- Mark Twain, "Pudd'nhead Wilson's Calendar"

Re: [PATCH net-next v2] bridge: also trigger RTM_NEWLINK when interface is released from bridge

[David Ahern]

On 9/20/17 3:09 PM, David Miller wrote:
> From: Vincent Bernat 
> Date: Sat, 16 Sep 2017 16:18:33 +0200
> 
> David, I am CC:'ing you because you've done work in this area over the
> past year.  I'm applying this patch, it's been sitting since the 16th
> and likes entirely correct to me.  But if you have objections just let
> me know.
> 
>> Currently, when an interface is released from a bridge via
>> ioctl(), we get a RTM_DELLINK event through netlink:
>>
>> Deleted 2: dummy0:  mtu 1500 master bridge0 
>> state UNKNOWN
>> link/ether 6e:23:c2:54:3a:b3
>>
>> Userspace has to interpret that as a removal from the bridge, not as a
>> complete removal of the interface. When an bridged interface is
>> completely removed, we get two events:
>>
>> Deleted 2: dummy0:  mtu 1500 master bridge0 state DOWN
>> link/ether 6e:23:c2:54:3a:b3
>> Deleted 2: dummy0:  mtu 1500 qdisc noop state DOWN group 
>> default
>> link/ether 6e:23:c2:54:3a:b3 brd ff:ff:ff:ff:ff:ff
>>
>> In constrast, when an interface is released from a bond, we get a
>> RTM_NEWLINK with only the new characteristics (no master):
>>
>> 3: dummy1:  mtu 1500 qdisc noqueue master 
>> bond0 state UNKNOWN group default
>> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
>> 3: dummy1:  mtu 1500 qdisc noqueue state 
>> UNKNOWN group default
>> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
>> 4: bond0:  mtu 1500 qdisc noqueue 
>> state UP group default
>> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
>> 3: dummy1:  mtu 1500 qdisc noqueue state DOWN group default
>> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
>> 3: dummy1:  mtu 1500 qdisc noqueue state DOWN group default
>> link/ether ca:c8:7b:66:f8:25 brd ff:ff:ff:ff:ff:ff
>> 4: bond0:  mtu 1500 qdisc noqueue 
>> state UP group default
>> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
>>
>> Userland may be confused by the fact we say a link is deleted while
>> its characteristics are only modified. A first solution would have
>> been to turn the RTM_DELLINK event in del_nbp() into a RTM_NEWLINK
>> event. However, maybe some piece of userland is relying on this
>> RTM_DELLINK to detect when a bridged interface is released. Instead,
>> we also emit a RTM_NEWLINK event once the interface is
>> released (without master info).
>>
>> Deleted 2: dummy0:  mtu 1500 master bridge0 
>> state UNKNOWN
>> link/ether 8a:bb:e7:94:b1:f8
>> 2: dummy0:  mtu 1500 qdisc noqueue state 
>> UNKNOWN group default
>> link/ether 8a:bb:e7:94:b1:f8 brd ff:ff:ff:ff:ff:ff
>>
>> This is done only when using ioctl(). When using Netlink, such an
>> event is already automatically emitted in do_setlink().

The DELLINK is for AF_BRIDGE family (ifi_family). Adding family to
print_linkinfo and running the monitor I get:


[LINK]family 0: 35: dummy1:  mtu 1500 qdisc
noqueue master br0 state UNKNOWN group default
link/ether d6:c3:73:86:3c:73 brd ff:ff:ff:ff:ff:ff

[LINK]family 7: 35: dummy1:  mtu 1500
master br0 state UNKNOWN
link/ether d6:c3:73:86:3c:73

[LINK]Deleted family 7: 35: dummy1:  mtu
1500 master br0 state UNKNOWN
link/ether d6:c3:73:86:3c:73

[LINK]family 0: 35: dummy1:  mtu 1500 qdisc
noqueue state UNKNOWN group default
link/ether d6:c3:73:86:3c:73 brd ff:ff:ff:ff:ff:ff

And that seems correct. So I think the RTM_NEWLINK added by this patch
should not be needed.

Re: [PATCH net-next v2] bridge: also trigger RTM_NEWLINK when interface is released from bridge

[David Miller]

From: Vincent Bernat 
Date: Sat, 16 Sep 2017 16:18:33 +0200

David, I am CC:'ing you because you've done work in this area over the
past year.  I'm applying this patch, it's been sitting since the 16th
and likes entirely correct to me.  But if you have objections just let
me know.

> Currently, when an interface is released from a bridge via
> ioctl(), we get a RTM_DELLINK event through netlink:
> 
> Deleted 2: dummy0:  mtu 1500 master bridge0 
> state UNKNOWN
> link/ether 6e:23:c2:54:3a:b3
> 
> Userspace has to interpret that as a removal from the bridge, not as a
> complete removal of the interface. When an bridged interface is
> completely removed, we get two events:
> 
> Deleted 2: dummy0:  mtu 1500 master bridge0 state DOWN
> link/ether 6e:23:c2:54:3a:b3
> Deleted 2: dummy0:  mtu 1500 qdisc noop state DOWN group 
> default
> link/ether 6e:23:c2:54:3a:b3 brd ff:ff:ff:ff:ff:ff
> 
> In constrast, when an interface is released from a bond, we get a
> RTM_NEWLINK with only the new characteristics (no master):
> 
> 3: dummy1:  mtu 1500 qdisc noqueue master 
> bond0 state UNKNOWN group default
> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
> 3: dummy1:  mtu 1500 qdisc noqueue state UNKNOWN 
> group default
> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
> 4: bond0:  mtu 1500 qdisc noqueue 
> state UP group default
> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
> 3: dummy1:  mtu 1500 qdisc noqueue state DOWN group default
> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
> 3: dummy1:  mtu 1500 qdisc noqueue state DOWN group default
> link/ether ca:c8:7b:66:f8:25 brd ff:ff:ff:ff:ff:ff
> 4: bond0:  mtu 1500 qdisc noqueue 
> state UP group default
> link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
> 
> Userland may be confused by the fact we say a link is deleted while
> its characteristics are only modified. A first solution would have
> been to turn the RTM_DELLINK event in del_nbp() into a RTM_NEWLINK
> event. However, maybe some piece of userland is relying on this
> RTM_DELLINK to detect when a bridged interface is released. Instead,
> we also emit a RTM_NEWLINK event once the interface is
> released (without master info).
> 
> Deleted 2: dummy0:  mtu 1500 master bridge0 
> state UNKNOWN
> link/ether 8a:bb:e7:94:b1:f8
> 2: dummy0:  mtu 1500 qdisc noqueue state UNKNOWN 
> group default
> link/ether 8a:bb:e7:94:b1:f8 brd ff:ff:ff:ff:ff:ff
> 
> This is done only when using ioctl(). When using Netlink, such an
> event is already automatically emitted in do_setlink().
> 
> Signed-off-by: Vincent Bernat 

Applied, thank you.

[PATCH net-next v2] bridge: also trigger RTM_NEWLINK when interface is released from bridge

[Vincent Bernat]

Currently, when an interface is released from a bridge via
ioctl(), we get a RTM_DELLINK event through netlink:

Deleted 2: dummy0:  mtu 1500 master bridge0 state 
UNKNOWN
link/ether 6e:23:c2:54:3a:b3

Userspace has to interpret that as a removal from the bridge, not as a
complete removal of the interface. When an bridged interface is
completely removed, we get two events:

Deleted 2: dummy0:  mtu 1500 master bridge0 state DOWN
link/ether 6e:23:c2:54:3a:b3
Deleted 2: dummy0:  mtu 1500 qdisc noop state DOWN group 
default
link/ether 6e:23:c2:54:3a:b3 brd ff:ff:ff:ff:ff:ff

In constrast, when an interface is released from a bond, we get a
RTM_NEWLINK with only the new characteristics (no master):

3: dummy1:  mtu 1500 qdisc noqueue master 
bond0 state UNKNOWN group default
link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
3: dummy1:  mtu 1500 qdisc noqueue state UNKNOWN 
group default
link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
4: bond0:  mtu 1500 qdisc noqueue state 
UP group default
link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
3: dummy1:  mtu 1500 qdisc noqueue state DOWN group default
link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff
3: dummy1:  mtu 1500 qdisc noqueue state DOWN group default
link/ether ca:c8:7b:66:f8:25 brd ff:ff:ff:ff:ff:ff
4: bond0:  mtu 1500 qdisc noqueue state 
UP group default
link/ether ae:dc:7a:8c:9a:3c brd ff:ff:ff:ff:ff:ff

Userland may be confused by the fact we say a link is deleted while
its characteristics are only modified. A first solution would have
been to turn the RTM_DELLINK event in del_nbp() into a RTM_NEWLINK
event. However, maybe some piece of userland is relying on this
RTM_DELLINK to detect when a bridged interface is released. Instead,
we also emit a RTM_NEWLINK event once the interface is
released (without master info).

Deleted 2: dummy0:  mtu 1500 master bridge0 state 
UNKNOWN
link/ether 8a:bb:e7:94:b1:f8
2: dummy0:  mtu 1500 qdisc noqueue state UNKNOWN 
group default
link/ether 8a:bb:e7:94:b1:f8 brd ff:ff:ff:ff:ff:ff

This is done only when using ioctl(). When using Netlink, such an
event is already automatically emitted in do_setlink().

Signed-off-by: Vincent Bernat 
---
 net/bridge/br_ioctl.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/net/bridge/br_ioctl.c b/net/bridge/br_ioctl.c
index 7970f8540cbb..3148cb3a8e82 100644
--- a/net/bridge/br_ioctl.c
+++ b/net/bridge/br_ioctl.c
@@ -99,8 +99,10 @@ static int add_del_if(struct net_bridge *br, int ifindex, 
int isadd)
 
if (isadd)
ret = br_add_if(br, dev);
-   else
+   else {
ret = br_del_if(br, dev);
+   rtmsg_ifinfo(RTM_NEWLINK, dev, IFF_MASTER, GFP_KERNEL);
+   }
 
return ret;
 }
-- 
2.14.1