Messages in this thread | | | Date | Fri, 11 Jun 1999 20:57:48 -0700 (PDT) | From | "John D. Hardin" <> | Subject | Re: [Masq-dev] Use daddr and dport as hash keys for faster masq redirector! |
| |
On Sat, 12 Jun 1999, Julian Anastasov wrote:
> The worse case will be if a masq module creates large number of rev > or normal masq entries with NO_DADDR and/or NO_DPORT flags to same > maddr and mport or to same saddr and sport and no connection take > place (flood from outside the firewall to create such inactive > entries).
And this case isn't too bad either - it's no worse than the current situation of not having a destination list.
> - we must use d{addr,port} as the only hash keys for > m_tab only when F_MPORT=1. So, better to use > d{addr,port}+m{addr,port} as keys for m_tab.
I disagree, still. The d{addr,port} lookup in __ip_masq_in_get() should find it quickly without having "special case" keying; this strikes me as an unnecessary complication.
> - using prev link for each table is a way to speedup > unhash/rehash for entries with value 0 in d{addr,port} > or s{addr,port},
Why would it hurt other entries? I think it's a win regardless.
I think you're trying to trade clarity and flexibility for memory footprint. I'd rather see the following lists:
s{addr,port} d{addr,port} m{addr,port} s+d{addr,port} m+d{addr,port}
...and no "special case" keying. For NO_D* | F_DLOOSE ip_masq_hash() could simply *not hash* the entry into the destination lists. This means 10 pointers per ms (each being a doubly-linked list), and two or five hash operations in ip_masq_hash() depending on the flags. Is this too expensive?
-- John Hardin KA7OHZ jhardin@wolfenet.com pgpk -a finger://gonzo.wolfenet.com/jhardin PGP key ID: 0x41EA94F5 PGP key fingerprint: A3 0C 5B C2 EF 0D 2C E5 E9 BF C8 33 A7 A9 CE 76 ----------------------------------------------------------------------- Efficiency can magnify good, but it magnifies evil just as well. So, we should not be surprised to find that modern electronic communication magnifies stupidity as *efficiently* as it magnifies intelligence. -- Robert A. Matern ----------------------------------------------------------------------- 90 days until 9/9/99
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |