diff options
| author | Patrick McHardy <kaber@trash.net> | 2007-12-18 07:45:52 +0100 |
|---|---|---|
| committer | David S. Miller <davem@davemloft.net> | 2008-01-28 23:59:11 +0100 |
| commit | 34498825cb9062192b77fa02dae672a4fe6eec70 (patch) | |
| tree | 7f856aff21ae3ab1328ab9bcf3549bca8e13df35 /net/ipv4 | |
| parent | [NETFILTER]: xt_hashlimit: reduce overhead without IPv6 (diff) | |
| download | linux-34498825cb9062192b77fa02dae672a4fe6eec70.tar.xz linux-34498825cb9062192b77fa02dae672a4fe6eec70.zip | |
[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>
Diffstat (limited to 'net/ipv4')
| -rw-r--r-- | net/ipv4/netfilter/ipt_CLUSTERIP.c | 2 | ||||
| -rw-r--r-- | net/ipv4/netfilter/nf_nat_core.c | 10 |
2 files changed, 8 insertions, 4 deletions
diff --git a/net/ipv4/netfilter/ipt_CLUSTERIP.c b/net/ipv4/netfilter/ipt_CLUSTERIP.c index a48e26449fd5..df39ca07fb12 100644 --- a/net/ipv4/netfilter/ipt_CLUSTERIP.c +++ b/net/ipv4/netfilter/ipt_CLUSTERIP.c @@ -273,7 +273,7 @@ clusterip_hashfn(const struct sk_buff *skb, } /* node numbers are 1..n, not 0..n */ - return (hashval % config->num_total_nodes) + 1; + return (((u64)hashval * config->num_total_nodes) >> 32) + 1; } static inline int diff --git a/net/ipv4/netfilter/nf_nat_core.c b/net/ipv4/netfilter/nf_nat_core.c index aec157d0ad93..e53ae1ef8f5e 100644 --- a/net/ipv4/netfilter/nf_nat_core.c +++ b/net/ipv4/netfilter/nf_nat_core.c @@ -77,10 +77,13 @@ EXPORT_SYMBOL_GPL(nf_nat_proto_put); static inline unsigned int hash_by_src(const struct nf_conntrack_tuple *tuple) { + unsigned int hash; + /* Original src, to ensure we map it consistently if poss. */ - return jhash_3words((__force u32)tuple->src.u3.ip, + hash = jhash_3words((__force u32)tuple->src.u3.ip, (__force u32)tuple->src.u.all, - tuple->dst.protonum, 0) % nf_nat_htable_size; + tuple->dst.protonum, 0); + return ((u64)hash * nf_nat_htable_size) >> 32; } /* Is this tuple already taken? (not by us) */ @@ -211,7 +214,8 @@ find_best_ips_proto(struct nf_conntrack_tuple *tuple, maxip = ntohl(range->max_ip); j = jhash_2words((__force u32)tuple->src.u3.ip, (__force u32)tuple->dst.u3.ip, 0); - *var_ipp = htonl(minip + j % (maxip - minip + 1)); + j = ((u64)j * (maxip - minip + 1)) >> 32; + *var_ipp = htonl(minip + j); } /* Manipulate the tuple into the range given. For NF_INET_POST_ROUTING, |