The lock-free algorithm has caused significant lookup performance regression for certain use cases. The regression is attributed to the use of non-relaxed memory orderings.
To address the issue, 2 versions of the lookup functions are created. One that uses the RW lock and the one that is lock-free. This restores the performance regression caused for use cases that used RW lock version of the lookup function. v2: 1) Adjusted the function call to take the lock in __rte_hash_lookup_with_hash_l (Jerin) 2) Squash all the intermediate commits into a single one Honnappa Nagarahalli (1): hash: separate lf and rw lock lookup code paths lib/librte_hash/rte_cuckoo_hash.c | 304 ++++++++++++++++++++++++++++-- 1 file changed, 290 insertions(+), 14 deletions(-) -- 2.17.1
