This is split out from the atomic deprecation series.
Convert lib/ring off rte_atomic32 and onto the C11 memory model,
except where C11 version has noticable performance drop
on x86 with GCC.
The pre-existing C11 and GCC-builtin paths lived in separate headers
with substantial duplication. After this series, only the MP head
CAS (__rte_ring_headtail_move_head_mt) retains separate implementations;
everything else is shared. Patch 2 documents the reason for
keeping the GCC builtin on the MP head CAS.
The default RTE_USE_C11_MEM_MODEL selection per architecture is unchanged.
Stephen Hemminger (5):
ring: split single thread vs multi-thread cases
ring: use GCC builtin as alternative to rte_atomic32
ring: use C11 for update_tail
ring: drop unused arg to update_tail
ring: use C11 for single thread move head
lib/ring/meson.build | 2 +-
lib/ring/rte_ring_c11_pvt.h | 61 +++------
lib/ring/rte_ring_elem_pvt.h | 116 ++++++++++++++++--
..._ring_generic_pvt.h => rte_ring_gcc_pvt.h} | 62 +++-------
lib/ring/rte_ring_hts_elem_pvt.h | 8 +-
lib/ring/soring.c | 34 ++---
6 files changed, 161 insertions(+), 122 deletions(-)
rename lib/ring/{rte_ring_generic_pvt.h => rte_ring_gcc_pvt.h} (63%)
--
2.53.0
