mbox series

[iproute2-next,0/6] bridge: vlan: add per-vlan options support

Message ID 20210418120137.2605522-1-razor@blackwall.org (mailing list archive)
Headers show
Series bridge: vlan: add per-vlan options support | expand

Message

Nikolay Aleksandrov April 18, 2021, 12:01 p.m. UTC
From: Nikolay Aleksandrov <nikolay@nvidia.com>

Hi,
This set extends the bridge vlan code to use the new vlan RTM calls
which allow to dump detailed per-port, per-vlan information and also to
manipulate the per-vlan options. It also allows to monitor any vlan
changes (add/del/option change). The rtm vlan dumps have an extensible
format which allows us to add new options and attributes easily, and
also to request the kernel to filter on different vlan information when
dumping. The new kernel dump code tries to use compressed vlan format as
much as possible (it includes netlink attributes for vlan start and
end) to reduce the number of generated messages and netlink traffic.
The iproute2 support is activated by using the "-d" flag when showing
vlan information, that will cause it to use the new rtm dump call and
get all the detailed information, if "-s" is also specified it will dump
per-vlan statistics as well. Obviously in that case the vlans cannot be
compressed. To change per-vlan options (currently only STP state is
supported) a new vlan command is added - "set". It can be used to set
options of bridge or port vlans and vlan ranges can be used, all of the
new vlan option code uses extack to show more understandable errors.
The set adds the first supported per-vlan option - STP state.
Man pages and usage information are updated accordingly.

Example:
 $ bridge -d vlan show
 port              vlan-id
 ens13             1 PVID Egress Untagged
                     state forwarding
 bridge            1 PVID Egress Untagged
                     state forwarding

 $ bridge vlan set vid 1 dev ens13 state blocking
 $ bridge -d vlan show
 port              vlan-id
 ens13             1 PVID Egress Untagged
                     state blocking
 bridge            1 PVID Egress Untagged
                     state forwarding

We plan to add many more per-vlan options in the future.

Thanks,
 Nik

Nikolay Aleksandrov (6):
  bridge: rename and export print_portstate
  bridge: add parse_stp_state helper
  bridge: vlan: add option set command and state option
  libnetlink: add bridge vlan dump request helper
  bridge: vlan: add support for the new rtm dump call
  bridge: monitor: add support for vlan monitoring

 bridge/br_common.h   |   3 +
 bridge/link.c        |  32 ++++--
 bridge/mdb.c         |   2 +-
 bridge/monitor.c     |  19 +++-
 bridge/vlan.c        | 255 +++++++++++++++++++++++++++++++++++++++++--
 include/libnetlink.h |   7 ++
 lib/libnetlink.c     |  19 ++++
 man/man8/bridge.8    |  75 ++++++++++++-
 8 files changed, 390 insertions(+), 22 deletions(-)

Comments

David Ahern April 22, 2021, 5:21 a.m. UTC | #1
On 4/18/21 5:01 AM, Nikolay Aleksandrov wrote:
> From: Nikolay Aleksandrov <nikolay@nvidia.com>
> 
> Hi,
> This set extends the bridge vlan code to use the new vlan RTM calls
> which allow to dump detailed per-port, per-vlan information and also to
> manipulate the per-vlan options. It also allows to monitor any vlan
> changes (add/del/option change). The rtm vlan dumps have an extensible
> format which allows us to add new options and attributes easily, and
> also to request the kernel to filter on different vlan information when
> dumping. The new kernel dump code tries to use compressed vlan format as
> much as possible (it includes netlink attributes for vlan start and
> end) to reduce the number of generated messages and netlink traffic.
> The iproute2 support is activated by using the "-d" flag when showing
> vlan information, that will cause it to use the new rtm dump call and
> get all the detailed information, if "-s" is also specified it will dump
> per-vlan statistics as well. Obviously in that case the vlans cannot be
> compressed. To change per-vlan options (currently only STP state is
> supported) a new vlan command is added - "set". It can be used to set
> options of bridge or port vlans and vlan ranges can be used, all of the
> new vlan option code uses extack to show more understandable errors.
> The set adds the first supported per-vlan option - STP state.
> Man pages and usage information are updated accordingly.
> 
> Example:
>  $ bridge -d vlan show
>  port              vlan-id
>  ens13             1 PVID Egress Untagged
>                      state forwarding
>  bridge            1 PVID Egress Untagged
>                      state forwarding
> 
>  $ bridge vlan set vid 1 dev ens13 state blocking
>  $ bridge -d vlan show
>  port              vlan-id
>  ens13             1 PVID Egress Untagged
>                      state blocking
>  bridge            1 PVID Egress Untagged
>                      state forwarding
> 
> We plan to add many more per-vlan options in the future.
> 

applied. Thanks, Nik