On Fri, Aug 10, 2012 at 09:42:10AM -0700, Ben Pfaff wrote: > On Thu, Aug 09, 2012 at 05:49:27PM +0900, Simon Horman wrote: > > Signed-off-by: Simon Horman <[email protected]> > > This patch is really quite odd. Before and after this patch, the > top-level function ofputil_decode_flow_stats_request() determines the > raw type and pulls off the OpenFlow header. But after this patch, the > function ofputil_decode_ofpst_flow_request() does all that work over > again. I think that you did cut-and-paste from > ofputil_decode_flow_stats_reply() into > ofputil_decode_ofpst_flow_request(), even though the situations are > different (a flow stats reply might have several flow stats > back-to-back, but a flow stats request has exactly one stats request).
My reasoning is that ofputil_decode_flow_stats_request() differentiates between OFP and NX requests while ofputil_decode_ofpst_flow_request() (is changed by a later patch so that it) differentiates between requests of different OF versions. I do, however, see that this involves some duplication of checking of the value of raw and will change things around so that. a) ofputil_decode_ofpst_flow_request is renamed ofputil_decode_ofpst_flow_request10() and otherwise unchanged from its current incarnation in the master brnach. b) I will add ofputil_decode_ofpst_flow_request11() and call it from ofputil_decode_flow_stats_request() c) ofputil_decode_nxst_flow_request() will remain unchanged _______________________________________________ dev mailing list [email protected] http://openvswitch.org/mailman/listinfo/dev
