I'm not sure why we wouldn't want to just deprecate the odp_packet_user_ptr/_u64() calls. They were stopgaps since we didn't have user metadata in v1.0. Now that we have that, what's the purpose of keeping the earlier calls? The latter is a more flexible and capable feature, and can be equally efficient for small user metadata areas (implementation choice).
On Wed, Apr 22, 2015 at 6:37 AM, Savolainen, Petri (Nokia - FI/Espoo) < [email protected]> wrote: > Hi, > > After thinking this again, it's cleaner for the API (and the user) if > these two user metadata fields are kept separate (as they currently are). > The user_ptr/_u64 is always present and user_data only when > param.pkt.udata_size is non-zero. Implementations have different ways to > store both of these into the packet descriptor (maybe pointer to user_data > is not needed, but it can be found with a fixed offset, etc). > > However the naming of these two could be cleaner (== highlight the > dependencies): > - odp_packet_user_ptr/_u64() could be renamed to > odp_packet_userctx_ptr/_u64() > - odp_packet_user_data/_data_size() could be renamed to > odp_packet_userarea/userarea_size() > - we should avoid confusion with similar terms in DPDK > - mbuf.userdata == current odp_packet_user_ptr() > - mbuf.udata64 == current odp_packet_user_u64() > > Opinions? I can send a patch for renames and documentation improvements. > > > -Petri > > > > > -----Original Message----- > > From: ext Zoltan Kiss [mailto:[email protected]] > > Sent: Tuesday, April 21, 2015 4:08 PM > > To: Bill Fischofer; [email protected]; Savolainen, Petri (Nokia - > > FI/Espoo) > > Subject: Re: [lng-odp] [API-NEXT PATCHv4 0/6] Add packet user metadata > > support > > > > It seems to me all questions were answered, is there anything which > > prevents applying this into master branch? > > Petri? > > > > On 10/04/15 16:52, Bill Fischofer wrote: > > > v4 changes: > > > - Removed RFC status, patch is now ready for API-NEXT > > > - Added implementation of approved APIs > > > - Added user metadata tests to packet validation test > > > > > > v3 changes: > > > - Renamed odp_packet_user_metadata() to odp_packet_user_data() > > > - Split addr/size return, adding odp_packet_user_data_size() > > > - Moved udata_size to pkt structure within odp_pool_param_t > > > > > > v2 changes: > > > - Moved udata_size to odp_pool_param_t > > > - Renamed odp_packet_udata() to odp_packet_user_metadata() > > > - Removed odp_buffer_udata(). User metadata is for packets only > > > > > > RFC for proposed minimal API set for user metadata support > > > based on today's discussions. Note that all initialization > > > and management of user metadata contents is the responsibility of > > > the ODP application. ODP APIs that copy system metadata will also > > > copy any associated user metadata as part of that operation, but > > > ODP will otherwise ignore these bytes. > > > > > > Bill Fischofer (6): > > > api: packet: add user metadata APIs > > > api: pool: add user metadata APIs > > > linux-generic: buffer: restructure user mdatadata fields > > > linux-generic: pool: add user metadata support > > > linux-generic: packet: add user metadata support > > > validation: packet: add user metadata tests > > > > > > include/odp/api/packet.h | 20 ++++++++ > > > include/odp/api/pool.h | 4 ++ > > > .../linux-generic/include/odp_buffer_internal.h | 4 +- > > > platform/linux-generic/include/odp_pool_internal.h | 2 +- > > > platform/linux-generic/odp_packet.c | 34 > ++++++++++++++ > > > platform/linux-generic/odp_pool.c | 34 > +++++--------- > > > test/validation/odp_packet.c | 54 > > ++++++++++++++++++++++ > > > 7 files changed, 126 insertions(+), 26 deletions(-) > > > >
_______________________________________________ lng-odp mailing list [email protected] https://lists.linaro.org/mailman/listinfo/lng-odp
