I don't think this is neccessary,cause you got the "on" keyword.
Filtering occurs in kernel AFTER the packet is passed into by the NIC
driver,
I think your mention would require seperate filters for each NIC, which
would
cause some things not to function properly (states etc).
Trust the skip-steps, the do what they should, well enough ( We have such a
bridge firewall here with about 400 rules, on a 400MHz PIII with 256MByte
RAM and NO load or latency problems, nor problems with throuput).
I think such a seperation (even only for config files) makes  no sense.

Did I get something wrong ?
enlight me if ...

----- Original Message -----
From: "Ed White" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Wednesday, February 05, 2003 4:24 PM
Subject: pf.conf and PF behaviour


> Hi,
>
> I've a quick question for PF developers:
>
> if PF checks ruleset everytime a packet pass through an interface this
means
> that for a classic gateway/bridge/firewall it will evaluate 2 times the
> ruleset. One going in if1 and going out if 2, right ?
>
> So Daniel have created skip-steps that let you jump all (or a lot of)
rules
> related to other interfaces.
>
> But why don't you separate ruleset files ?
>
> pf.conf (all global definitions)
> pf.rl0
> pf.fxp0
> pf.dc0
> pf.dc1
> pf.tun0
>
> So you'll be sure to evaluate interface related rules only.
>
> What about ?
>
>
> Ed
>
>
>
>
>


Reply via email to