[NETFILTER]: non-power-of-two jhash optimizations
Apply Eric Dumazet's jhash optimizations where applicable. Quoting Eric:
Thanks to jhash, hash value uses full 32 bits. Instead of returning
hash % size (implying a divide) we return the high 32 bits of the
(hash * size) that will give results between [0 and size-1] and same
hash distribution.
On most cpus, a multiply is less expensive than a divide, by an order
of magnitude.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/net/netfilter/nf_conntrack_expect.c b/net/netfilter/nf_conntrack_expect.c
index 0efbf34..e0cd9d0 100644
--- a/net/netfilter/nf_conntrack_expect.c
+++ b/net/netfilter/nf_conntrack_expect.c
@@ -73,15 +73,17 @@
static unsigned int nf_ct_expect_dst_hash(const struct nf_conntrack_tuple *tuple)
{
+ unsigned int hash;
+
if (unlikely(!nf_ct_expect_hash_rnd_initted)) {
get_random_bytes(&nf_ct_expect_hash_rnd, 4);
nf_ct_expect_hash_rnd_initted = 1;
}
- return jhash2(tuple->dst.u3.all, ARRAY_SIZE(tuple->dst.u3.all),
+ hash = jhash2(tuple->dst.u3.all, ARRAY_SIZE(tuple->dst.u3.all),
(((tuple->dst.protonum ^ tuple->src.l3num) << 16) |
- (__force __u16)tuple->dst.u.all) ^ nf_ct_expect_hash_rnd) %
- nf_ct_expect_hsize;
+ (__force __u16)tuple->dst.u.all) ^ nf_ct_expect_hash_rnd);
+ return ((u64)hash * nf_ct_expect_hsize) >> 32;
}
struct nf_conntrack_expect *