Fri, May 08, 2015 at 04:19:15PM CEST, john.r.fastabend@intel.com wrote:

On 05/07/2015 09:10 AM, Jiri Pirko wrote:

quoted

Tue, May 05, 2015 at 07:34:54PM CEST, sfeldma@gmail.com wrote:

quoted

From: Scott Feldman <redacted>

The IPv4 FIB ops convert nicely to the switchdev objs and we're left with
only four switchdev ops: port get/set and port add/del.  Other objs will
follow, such as FDB.  So go ahead and convert IPv4 FIB over to switchdev
obj for consistency, anticipating more objs to come.

Signed-off-by: Scott Feldman <redacted>

Acked-by: Jiri Pirko <jiri@resnulli.us>

Hi Scott, Jiri,

Now that your converting IPv4 and FDB over to objs this starts to look like
table operations to me. You have add/del/dump operations on them but are
missing the resource piece and how objects are inter-related. I assume in
the end you will have an "obj" for each one of your hardware tables.

Yes, these are objects, but very well defined ones. FDB entry is well
defined, I don't see a need to abstract it.

At this point I think it might be better to lay this onto the Flow API
I wrote but tighten up the Flow API logic a bit to constrain it. This
patch at least doesn't appear to leak into userspace at all which is good
so there is no problem doing the conversion after this is applied. This
will help I think when Roopa adds the improved failure schemes. Because
at least in some of my use cases I want to have some sense when rules
are going to start failing and how they are interconnected with other
tables before I start working on the switch.

I need to see that patch you mention. But anyway, we can certainly
change the lower parts when needed.

Thanks,
.John

quoted

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html