On 03/26/14 01:37, Roopa Prabhu wrote:

On 3/25/14, 1:11 PM, Florian Fainelli wrote:

quoted

2014-03-25 12:35 GMT-07:00 Neil Horman [off-list ref]:

Sorry about getting on this thread late and possibly in the middle.
Agree on the idea of keeping the ports linked to the master switch dev
(or the 'conduit' to the switch chip) via private list instead of the
master-slave relationship proposed earlier.
By private i mean the netdev->priv linkage to the master switch dev and
not really keeping the ports from being exposed to the user.

We think its better to keep the switch ports exposed as any other netdev
on linux.
  This approach will make the switch ports look exactly like a nic port
and all tools will continue to work seamlessly. The switch port
operations could internally be forwarded to the switch netdev (sw1 in
the above case).

example:
$ip link set dev sw1p0 up
$ethtool -S sw1p0

I like the approach. I know the above is a simple version, but i am
assuming you also mean i can do things like
ip route add ...
bridge fdb add ... (and if you like your brctl go ahead)
bonding ...

whether sw1 is needed as a separate netdev existing on the system is
debatable.

I dont see need to expose it. For 1, it will be confusing to have this
netdev whose only task is to control the chip.

Most cases the switch port driver (API) can talk to the switch chip
driver without a switch netdev in between.
But there are cases where a switch netdev might become necessary for
switch chip specific operations (which probably has been discussed on
this thread). An example could be a global acl rule that applies to all
switch ports. One can argue that this can be applied on individual
switch ports and the switch driver can take care of consolidating or
optimally programming the acl rule in the switch chip.

There are a lot of things which dont tie to a specific port.
I think these should be transparent to the chip. If i add a route
and the decision is for that route to go to the chip, then it
shows up at the driver and it goes to the ASIC.
If i am dumping a fib table and some parts of it sit in the chip,
then whatever interfaces would end up querying the chip.

cheers,
jamal

Thanks,
Roopa