From owner-freebsd-hackers  Sat Jun 24 14:13:39 1995
Return-Path: hackers-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id OAA25815
          for hackers-outgoing; Sat, 24 Jun 1995 14:13:39 -0700
Received: from Root.COM (implode.Root.COM [198.145.90.1])
          by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id OAA25808
          for <hackers@freebsd.org>; Sat, 24 Jun 1995 14:13:34 -0700
Received: from corbin.Root.COM (corbin [198.145.90.18]) by Root.COM (8.6.11/8.6.5) with ESMTP id OAA00885; Sat, 24 Jun 1995 14:13:35 -0700
Received: from localhost (localhost [127.0.0.1]) by corbin.Root.COM (8.6.11/8.6.5) with SMTP id OAA00135; Sat, 24 Jun 1995 14:13:58 -0700
Message-Id: <199506242113.OAA00135@corbin.Root.COM>
To: Network Coordinator <nc@ai.net>
cc: hackers@freebsd.org
Subject: Re: FreeBSD as a router 
In-reply-to: Your message of "Sat, 24 Jun 95 17:05:29 EDT."
             <Pine.BSF.3.91.950624170016.2548B-100000@aries.ai.net> 
From: David Greenman <davidg@Root.COM>
Reply-To: davidg@Root.COM
Date: Sat, 24 Jun 1995 14:13:58 -0700
Sender: hackers-owner@freebsd.org
Precedence: bulk

>So the problem is how to get BSD to handle packets faster. And I guess my 
>question still is, why can't we move some of the packet-handling and routing 
>directly into the driver where it is a few layers closer to the actual 
>hardware. Not really off loading, but giving the packet-handling code a 
>chunk of the CPU time (probably as much as it needs) w/o being able to be 
>squeezed out by other processes and such. 

   If there is a way that can be found to do it in an architecturally clean
way, then we may very do this...but is may not be necessary. I haven't had a
chance to look carefully at where CPU is being spent when routing packets. On
of these days...

>There is a program called pc-route for DOS systems that supposedly is as 
>fat-free as code can be [no branches in the assembly source, etc]. On a 
>pentium with 2 100 mbps cards, I am wondering how fast it could move 
>packets to give a theoretical packet/s maximum. Anyone have a 
>configuration where they could try it?

   Last time I used pc-route, it crashed every 5-30 minutes. It's performance
wasn't so hot, either. I haven't looked at it in a year or so, so perhaps the
code has been improved. The main things that stick in my memory are that it was
a black box, difficult to configure, impossible to troubleshoot, and had a
broken RIP implementation.

-DG