> On Nov 16, 2017, at 1:37 PM, Olivier MATZ <olivier.m...@6wind.com> wrote: > > On Thu, Nov 16, 2017 at 01:32:13PM +0400, Ilya Matveychikov wrote: >> >>> On Nov 16, 2017, at 1:06 PM, Hanoch Haim (hhaim) <hh...@cisco.com> wrote: >>> >>> Understood >>> >>> rte_mbuf_refcnt_update_blind() >>> >>> should be good., it will take care the RTE_MBUF_REFCNT_ATOMIC >>> >> >> >> Why guys not to add just __rte_mbuf_refcnt_update() as a wrapper over >> rte_atomic16_add_return() and use it in inside rte_mbuf_refcnt_update() and >> rte_pktmbuf_prefree_seg() as well? >> > > Is there any other difference with rte_mbuf_refcnt_update_blind() except > the function name?
No really, but my suggestion was not only about the name but to use such a function inside rte_mbuf_refcnt_update() too. Also, that is common naming scheme in Linux kernel — to add “__” prefix for for “lightweight” functions. Anyway, IMO having a function will be better than having ifdef/else/endif block.
