Mike Christie wrote: >> 3. In think if we implement above two then additional exch ref for abort is >> not required and fcp would ensure by it fsp state to call fc_exch_done() >> only once to free the exch. However if you think for some corner case this >> is still required then fc_exch.c should only increment and decrement that >> exch ref for abort, increment on abort issue and decrement when abort >> handling is done. >> > > Taking a refcount is nice because it is consistent with other operations > where we send a sequence or even another exhange that references some > object. If we are saying we are going to sync evreything up in fc_exch.c > like below then this is a special case so who cares about consistency > for this case :) >
Actually, I like to keep it consistent. It is weird that we would take a ref for the rrq, but not take one for the abort. It is easier to review if we do not add special cases all around. Hey also on the subject of rrq if we send a rrq from fc_exch_rrq wouldn't we call fc_exch_rrq_resp for the response, and if so shouldn't it drop the hold taken in fc_exch_abts_resp. _______________________________________________ devel mailing list [email protected] http://www.open-fcoe.org/mailman/listinfo/devel
