[ovs-dev] [ovs-controller 3/3] ovs-controller: Allow --wildcards to specify a wildcard set.

[Ben Pfaff]

On Wed, Jun 01, 2011 at 01:10:02AM -0700, Justin Pettit wrote:
> On May 31, 2011, at 4:15 PM, Ben Pfaff wrote:
> 
> > +    if (cfg->wildcards) {
> > +        uint32_t ofpfw;
> > +
> > +        if (cfg->wildcards == UINT32_MAX) {
> > +            /* No wildcard defined, try to wildcard as many fields
> > +             * as possible, however we cannot wildcard all fields.
> > +             * We need in_port to detect moves.
> > +             * We need both MAC SA and DA to do L2 learning. */
> 
> Do you want to mention that we want VLAN VID to do L2 learning, too?

I rephrased this as:

	/* Try to wildcard as many fields as possible, but we cannot
	 * wildcard all fields.  We need in_port to detect moves.  We need
	 * Ethernet source and dest and VLAN to do L2 learning. */


> > +The optional \fIwildcard_mask\fR is an OpenFlow wildcard bitmask in
> > +hexadecimal that specifies the fields to wildcard.  If no
> > +\fIwildcard_mask\fR is specified, the default value 0x2820F0 is used
> > +which specifies L2-only switching and wildcards L3 and L4 fields.
> > +Another interesting value is 0x2000EC, which specifies L3-only
> > +switching and wildcards L2 and L4 fields.
> 
> Apple's got a lot of cash...it's amazing they haven't swooped us up
> yet for their interface team.

This interface is terrible but ovs-controller is really just for
debugging.  I hope no one actually uses it in production for anything.

> > -/* -w, --wildcard: Set up exact match or wildcard flow entries? */
> > -static bool exact_flows = true;
> > +/* -w, --wildcard: 0 to disable wildcard flow entries, a OFPFW_* bitmask to
> > + * enable wildcards, or UINT32_MAX to use the default wildcards. */
> 
> Maybe change "enable wildcards" to "enable specific wildcards"?

Done.

I'll push this soon.