Tue, Mar 31, 2015 at 02:08:34AM CEST, ronen.arad@intel.com wrote:

quoted

-----Original Message-----
From: Scott Feldman [mailto:sfeldma@gmail.com]
Sent: Monday, March 30, 2015 2:28 PM
To: Arad, Ronen
Cc: roopa; Netdev; Jirí Pírko; Guenter Roeck; Florian Fainelli
Subject: Re: [PATCH net-next 11/18] switchdev: remove old
netdev_switch_port_bridge_setlink

On Mon, Mar 30, 2015 at 1:46 PM, Arad, Ronen [off-list ref] wrote:

quoted

quoted

-----Original Message-----
From: netdev-owner@vger.kernel.org [mailto:netdev-owner@vger.kernel.org] On
Behalf Of Scott Feldman
Sent: Monday, March 30, 2015 1:20 PM
To: roopa
Cc: Netdev; Jiří Pírko; Guenter Roeck; Florian Fainelli
Subject: Re: [PATCH net-next 11/18] switchdev: remove old
netdev_switch_port_bridge_setlink

On Mon, Mar 30, 2015 at 6:23 AM, roopa [off-list ref] wrote:

quoted

On 3/30/15, 1:40 AM, sfeldma@gmail.com wrote:

quoted

From: Scott Feldman <redacted>

New attr-based bridge_setlink can recurse lower devs and recover on err,
so
remove old wrapper.  Also, restore br_setlink back to original and don't
call
into SELF port driver.  rtnetlink.c:bridge_setlink already does a call
into
port driver for SELF.

Signed-off-by: Scott Feldman <redacted>

removing this now requires every vlan add to be a two step process, why ?

No, that's not true.  You want to use
ndo_vlan_rx_add_vid/ndo_vlan_rx_kill_vid in your port driver, and then
using either vlan driver standalone or the bridge driver vlan support
will work.

Try it.  Implement ndo_vlan_rx_add_vid in your port driver and verify
you get called to add VLAN to port with either:

   bridge vlan add dev swp1 vid 10

   -or-

   vconfig add swp1 10

Same for deleting a VLAN, either of these two commands call into the
port driver ndo_vlan_rx_kill_vid:

   bridge vlan del dev swp1 vid 10

   -or-

   vconfig rem swp1 10

quoted

bridge vlan add dev swp1 vid 10
bridge vlan add dev swp1 vid 10 self

Not necessary.  The first command is sufficient if using
ndo_vlan_rx_add_vid.

This is not sufficient for VLAN filtering. Ndo_vlan_rx_add_vid does not
provide the vinfo flags PVID and UNTAGGED. Therefore it is not
an adequate replacement for propagating setlink/dellink messages to the
swithport driver or an alternative via swdev_attr.

Glad you bring that point up.  I think these can get cast as port
attrs and set using swdev_attr.  This is something swdev attr should
open up is allowing more settings to be pushed down to port driver.
I'll look into this one and include it with v2.

It could be beneficial to build extensibility into swdev_attr.
An experimenter attribute designed to carry arbitrary data could allow
for passing new attributes and implementation specific attributes
without affecting any existing switchdev driver:

Warning sign...

I believe that we don't want this. It is very easy to add attribute for
anything. Having this "universal attribuute" only allows wild things...

Thanks.

Jiri

enum swdev_attr_id {
	SWDEV_ATTR_UNDEFINED,
SWDEV_ATTR_EXPERIMENTER,
	SWDEV_ATTR_PORT_PARENT_ID,
SWDEV_ATTR_PORT_STP_STATE,
};


struct swdev_experimenter_attr {
u32 exp_id; 	/* - MSB 0: low-order bytes are IEEE OUI */
		/* - MSB != 0: reserved for netdev */
u32 exp_attr;
u16 exp_attr_size;
const void *exp_attr_data;
}

struct swdev_attr {
enum swdev_attr_id attr;
u32 flags;

	union {
		struct netdev_phys_item_id ppid;	/* PORT_PARENT_ID */
		u8 stp_state;				/* PORT_STP_STATE */
	unsigned long brport_flags;	/* PORT_BRIDGE_FLAGS */
	/* netdev defined attributes abobe this line */
	struct swdev_experimenter_attr exp_attr;
	};