Hi, talking with SiegeX6 on IRC we found consensus that the struct xt_conntrack_mtinfo1 is just too fat -- 88 bytes if I counted right. 64 of that go away for supporting IPv6 masking, which is plenty. We could use a uint8_t CIDR field instead of 'union nf_inet_addr origsrc_mask', and use a lookup table: static const struct { union nf_inet_addr expanded; unsigned char contracted; } table[] = { {IN6_ADDR(0000,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0), 0}, {IN6_ADDR(8000,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0), 1}, {IN6_ADDR(c000,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0), 2}, {IN6_ADDR(e000,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0), 3}, {IN6_ADDR(f000,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0), 4}, {IN6_ADDR(f800,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0), 5}, /* and so on */ }; This would cost us 2048 bytes once. Everything that uses IPv6 CIDR<->mask transformation could use this. - xt_conntrack: save 60 bytes per struct - xt_hashlimit: save on some static computation power (currently, xt_hashlimit computes the mask from CIDR during rule insertion) - xt_connlimit: save 15 bytes per struct (realistically: 12, due to aligned(8) padding) - xt_policy: save 30 bytes per struct (realistically 24) - ipt_entry, ip6t_entry: basically, these too, but it would touch a non-revisionable structure - can't break it - probably tons of other code in non-netfilter areas in net/ Are there any objections to having this big table? - To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html