On Fri, May 10, 2013 at 08:23:10AM +0900, Simon Horman wrote: > On Thu, May 09, 2013 at 09:22:07AM -0700, Ben Pfaff wrote: > > On Thu, May 02, 2013 at 06:06:46PM +0900, Simon Horman wrote: > > > Update handling of the following actions to use the dl_type set by MPLS > > > push and pop actions if it differs from the original dl_type. This is > > > consistent with the existing checking of load actions and allows > > > their existing checks to enforce dl_type pre-requisites correctly. > > > > > > output_reg > > > bundle > > > reg_move > > > stack_push > > > stack_pop > > > learn > > > multipath > > > > > > In order to avoid the verbosity of updating the flow for each applicable > > > action the update is treated as a common case and performed at the top of > > > ofpact_check__(). This was suggested by Jesse Gross. > > > > > > Cc: Jesse Gross <je...@nicira.com> > > > Signed-off-by: Simon Horman <ho...@verge.net.au> > > > > Jesse, it looks like you suggested this, do you want to review it? > > > > I think that there may be a related issue though: it looks to me like > > a pop_mpls action has no effect on dl_type if there is no MPLS header, > > but I think that the logic for checking actions assumes that pop_mpls > > always updates dl_type. > > Thanks, I'll look into that.
It seems to me that a way to resolve this would be to restrict pop_mpls by making an MPLS dl_type a pre-requisite in ofpact_check__(). However, this would be a somewhat of a departure from the checking of a many of other actions. The example Jesse gave me recently was that setting the TCP port is permitted on a completely wildcarded flow. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
