[ovs-dev] [PATCH v2 2/4] ovn: Implement the ability to send a packet back out its input port.

Ben Pfaff blp at nicira.com
Sun Oct 18 19:49:17 UTC 2015


On Sun, Oct 18, 2015 at 10:58:35AM -0700, Justin Pettit wrote:
> 
> > On Oct 17, 2015, at 2:07 PM, Ben Pfaff <blp at nicira.com> wrote:
> > 
> > 
> > +            if (!port && sf->field->id == MFF_REG6) {
> 
> Is there a reason you're using MFF_REG6 instead of MFF_IN_PORT?

MFF_LOG_INPORT, which is what I want, wasn't available here.  We've run
into this problem before.  I'm going to propose moving the MFF_LOG_*
definitions into a new header.  I'll post a v3 of the series that does
that.

> > +                sf = ofpact_put_SET_FIELD(ofpacts);
> > +                sf->field = mf_from_id(MFF_IN_PORT);
> > +                bitwise_put(UINT64_MAX, &sf->mask, sf->field->n_bytes, 0,
> > +                            sf->field->n_bits);
> 
> Is there a reason to use this instead of bitwise_one(), which seems to
> have clearer intent?  The code right above doesn't do it either, so
> I'm just curious if it's not equivalent.

bitwise_one() works nicely too.  Previously, I guess I used
bitwise_put() because it had the same form as the bitwise_put() in the
pairing of value and mask:

            bitwise_put(port, &sf->value,
                        sf->field->n_bytes, 0, sf->field->n_bits);
            bitwise_put(UINT64_MAX, &sf->mask,
                        sf->field->n_bytes, 0, sf->field->n_bits);

and then in this case I just cut-and-paste the code.

I'll change them both to bitwise_one().

> Acked-by: Justin Pettit <jpettit at nicira.com>

Thanks, I'll post v3 in a few minutes.



More information about the dev mailing list