3.1.2 Kernel-requirements

Here are the necessary changes for `.../conf/MACHINE' in order to incorporate the various networking-facilities into your kernel:

• Enable TCP/IP networking in general:

  option         INET             # Basic networking support - mandatory
  pseudo-device  loop	1       # network loopback - mandatory
  

• Put this entry in if you want to run SLIP:

  pseudo-device  sl      1       # Serial Line IP (SLIP)
  

• This one is needed to do PPP:

  pseudo-device  ppp     1       # Point-to-Point-Protocol (PPP)
  

• If you're proud owner of an Ethernet-card, add any of the following items:

  le0		at zbus0		# A2065, Ameristar and ASDG (Lance) ethernet. 
  ed0		at zbus0		# Hydra (dp8390) ethernet
  es0		at zbus0		# A4066 (SMC 91C90) ethernet
  qn0		at zbus0		# Quicknet ethernet
  ae0		at zbus0		# Ariadne ethernet
  

• to get Arcnet-support, add the following lines:

  bah0		at zbus0		# C= arcnet
  

• If you want to use the two ports of a MultiFaceCard, use these settings:

  mfc0		at zbus0		# MultiFaceCard I/O board
  mfcs0		at mfc0 unit 0		# MFC serial
  mfcs1		at mfc0 unit 1		# MFC serial
  

• Enable the following two options, if you want to run your system as NFS-client or -server:

  options		NFSSERVER	# Network File System server side code
  options		NFSCLIENT	# Network File System client side code
  

• It your system has multiple network interfaces and you want it to route between them, add the following line to get the kernel do it:

  options	GATEWAY		# Packet forwarding
  

  Please note that you can switch this on/off on a running system with sysctl -w net.inet.ip.forwarding=1/0.
• Usually, broadcasts (packets sent to the broadcast address) are not forwarded outside the local subnet. To change this behaviour, add the following line:

  options	DIRECTED_BROADCAST	# Broadcast across subnets
  

• If you want to enable multicast routing (used for video conferences, etc.), do this with the this option:

  options	MROUTING	# Multicast routing
  

• Just for the sake of completenes, here's how to enable any of the non-IP-based protocols supportet by NetBSD:

  options	NS		# XNS
  options	ISO,TPIP	# OSI
  options	EON		# OSI tunneling over IP
  options	CCITT,LLC,HDLC	# X.25
  

• If you want to get the old 4.2BSD's behaviour regarding broadcasts, add the following option to your config file:

  options		"TCP_COMPAT_42"	# Use 4.2 BSD style TCP
  

• It's possible to read packets routed to a certain networking device (tunX) from a device (/dev/tunX), and to route data written to the same device as if it came from that networking device. With this tunneling mechanism, one can e.g. implement dial-on-demand on for SLIP/PPP (NetBSD V1.2's pppd does DOD without hacking the tunnel-device), etc. The size of the packets read from/written to the /device is the same as given to ifconfig by the mtu value. To enable the tunneling mechanism, add the following line:

  pseudo-device	tun	1		# network tunnel line discipline
  

• Certain utilities & daemons need access to the packets sent on the network, among them are tcpdump and rarpd. The packets are intercepted using the Berkeley Packet Filter interface:

  pseudo-device	bpfilter	4	# berkeley packet filters
  

Most of these options are already included in the GENERIC kernel. They are listed here rather as an hint for those who want to known what to exclude from a kernel. Leaving out all the networking stuff should save you about 300-500k of memory, but please note that e.g. X needs TCP/IP networking facilities.