On Wed, Jun 26, 2013 at 12:42:48PM -0700, Ben Pfaff wrote:
> On Mon, Jun 24, 2013 at 06:59:15PM -0700, Ethan Jackson wrote:
> > Before this patch, the mirroring code would retroactively insert
> > actions for destination mirrors after actions were translated.
> > This relied on converting datapath output actions into ofports
> > which doesn't work for tunnels and patch ports. This patch
> > refactors the code to handle destination mirrors at output.
> >
> > Signed-off-by: Ethan Jackson <et...@nicira.com>
>
> The approach here seems odd, so I wonder if I'm missing something.
>
> What I expected to see was that each compose_output_action() would add
> the mirrors for that output to a set of mirrors to ultimately use, and
> that somewhere else (just one place) we would add the mirrors for the
> input to this same set, and that at the end we would use the set of
> mirrors to do the mirroring.
>
> What I see here is that each compose_output_action() computes the set
> of input mirrors, adds in the provided set of output mirrors,
> subtracts the mirrors already output to, and then outputs to any
> remaining ones.
>
> Am I missing something?
Oh, one more thing. The formatting here implies that 'orig_flow' is
related to "Flow at the last commit." I don't think it really is:
/* Flow at the last commit. */
struct flow base_flow;
struct flow orig_flow; /* Junk unless 'hit_resubmit_limit' or mirroring. */
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
