On Tue, Oct 12, 2021 at 9:41 PM Alexei Starovoitov <ast@xxxxxx> wrote: > > On 10/12/21 5:11 PM, Martin KaFai Lau wrote: > > On Tue, Oct 12, 2021 at 03:46:47PM -0700, Joanne Koong wrote: > >> I'm also open to adding the bloom filter map and then in the > >> future, if/when there is a need for the bitset map, adding that as a > >> separate map. In that case, we could have the bitset map take in > >> both key and value where key = the bitset index and value = 0 or 1. > > v4 uses nr_hash_funcs is 0 (i.e. map_extra is 0) to mean bitset > > and non-zero to mean bloom filter. > > > > Since the existing no-key API can be reused and work fine with bloom > > filter, my current thinking is it can start with bloom filter first > > and limit the nr_hash_funcs to be non-zero. > > > > If in the future a bitset map is needed and the bloom filter API can > > be reused, the non-zero check can be relaxed. If not, a separate > > API/map needs to be created for bitset anyway. > > > > sounds good to me. > let's drop bitset for now since there doesn't seem to be a consensus. sgtm too
