Gilles Lamiral,

$Revision: 1.40 $ $Date: 2002/07/07 23:46:32 $

This paper describes how to install a Gnu|Linux distribution on a computer without Ethernet card, or CD-ROM, but just a local floppy drive and a remote nfs server attached by a Null-Modem parallel cable. The nfs server has a cdrom drive mounted and exported.

You can find the latest release of the PLIP-Install-HOWTO, by Gilles Lamiral, located at:

If you have already read this HOWTO and you wonder what is new in this release, first read the section what's new.

1. PLIP for the impatient

2. Introduction

3. License

4. How did I write this howto ?

5. What do you need ?

6. The documentation

7. Conventions

8. Building your own cable

9. Network parameters

10. PLIP on the Source side

11. The cdrom nfs server side

12. FIPS, PARTED, splitting of hard disk partitions

13. Debian installation

14. Installing Slackware, Caldera

15. Installing RedHat 6.x

16. Installing RedHat 7.0

17. Installing Suse

18. Installing from a DOS partition

19. Install the plip interface permanently

20. Compatibility with a printer (lp)

21. Need to compile a new kernel ?

22. Known problems

23. What's new

24. To do

25. List of contributors and acknowledgements

1. PLIP for the impatient

This section is used to avoid reading the document when you need to install a plip connection in one minute. Skip it.

1.1 Source side

cat /proc/parport/0/hardware
insmod parport
insmod parport_pc io=0x378 irq=7 # Got those values in BIOS setup.
insmod plip
tail /var/log/messages
ifconfig plip0 pointopoint \
         netmask up
ifconfig plip0
ping -c1
route -n

1.2 Target side

ifconfig plip1  pointopoint \
         netmask up
route add -host dev plip1 # if the route is not present

1.3 Final tests

ping # from
ping # from

Now all TCP/IP is available between the two computer via the parallel port. Have a nice trip !

2. Introduction

"PLIP is Parallel Line IP, that is, the transportation of IP packages over a parallel port." Donald Becker (

Writing a PLIP install HOWTO seems useless nowadays since Ethernet cards are cheap: NE2000 cards cost about the same as a Null-Modem cable. This is true for desktop computers, but not for laptop computers, in which the PCMCIA card costs more than 10 times as much the Null-Modem cable. Moreover, there is a parallel port on every computer, but not always a network card.

Of course, this howto can be used to install Linux on every personal computer without loss of generality.

This paper is just what I've done to install a Debian GNU/Linux distribution on a Toshiba Portégé 620CT laptop, from a NFS exported cdrom drive, via a Null-Modem cable.

A Null-Modem cable is also called a Null-Printer cable.

A Null-Modem cable is also called a "LapLink" (or "Turbo Laplink") cable but the word "LapLink" is trademarked by Traveling Software under the number 75466713 since 1986, so I won't use it anymore.

This HOWTO will be obsolete when every Linux distribution includes a PLIP install option. For example, the Debian installation only needs to add two commands to make this HOWTO obsolete (ifconfig + route). I hope one Debian maintainer will consider this point.

I hoped this HOWTO would become obsolete but in fact it seems that it will never be, because some builders of some distributions do include PLIP installation sometimes but forget to maintain it in new releases (REDHAT for example).

I would be happy to know if someone used the PLIP-Install-HOWTO to install other Linux distributions from other network protocols (ftp, http, nfs, samba, or even NT/Novell servers)

Feedback with typos, bad English, comments, money, job, joy, fears, cries are welcome and recommended (not all with the same eagerness).

If a translation in any language exists, please let me know, I'll be happy.

3. License

This documentation used to be under the OpenContent license. It is not anymore. Do what you want with this document.

4. How did I write this howto ?

I wrote this paper during the installation process but since I encountered several problems at the same time, my notes were disparate and sometimes I forgot to write the good command lines because I was too fed up to reboot (kernel compilations) and edit this file each time.

When the installation went wrong (it did, I'm a beginner on laptops, plip, linux nfs, portmapper, Debian), I tried several things to fix the problems. When they remained too long, my brain was focused on solving, not writing.

So, when the installation process finally succeeded, I decided to restart the process from scratch by noting everything in order to save your time. It was the first time I had installed Linux twice on the same computer.

Then, in order to verify the document, I again restarted from scratch with a Slink Debian distribution (the next after the Hamm one). It was the first time I had installed Linux 3 times on the same computer. I hope it was the last time I have to do that.

Conclusion: If you do what is in this paper, it should work.

5. What do you need ?

6. The documentation

I recommend these good readings in case you can't figure out how to solve a problem. I've read them. You should too if you're curious or conscientious.

Please, do not forget to consider the LDP mirrors, listed at:

Most HOWTOs are translated in many languages. Just go to the bottom of a mirror page and follow the translations/ link.

If you want to use Linux on a laptop, read the latest Linux-Mobile-Guide, by Werner Heuser, located at:

If you haven't installed any distribution yet, carefully read the Installation-HOWTO, by Eric S. Raymond, located at: And make more coffee :-)

If you need information about your cdrom drive, read the CDROM-HOWTO, by Jeff Tranter, located at:

If you haven't installed an nfs server yet, read the NFS-HOWTO, by Nicolai Langfeldt, located at:

If you need to compile a new kernel on the source box, read the Kernel-HOWTO, by Brian Ward, located at:

If you are new to PLIP, read the PLIP MINI-HOWTO, by Andrea Controzzi, located at:

If you plan to make your Null-Modem cable yourself, a good reading is the file PLIP.txt, by Donald Becker, located at:
on your source linux computer, Luke. Alternatively, read my notes in section "Building your own cable".

7. Conventions

The file contents and line commands and install screen-shots are always in typewriter font, like this:

     #### This is the great file /bin/Windows ####
     while [ "1" ]; do
             echo "I do my best because I'm the best"
             echo "Very soon, next Y2Kill (the 01/01/0000)"
             echo "A new marvelous 64 bit release !"
             echo "Please wait a little more"
             sleep 18446744073709551615 # 2^64-1

or this:

     $ killall Windows

The file content lines should never begin with white space. You'll have to remove them, if any. Sorry, I'm fed up with C-a M-AltGr-\ (remember, I'm a French azerty writer). Tab-emacs reflex is untameable (coders who use Emacs always press the tabular key like a twitch).

Command input lines begin with a dollar $ (the prompt), you don't have to type the dollar, just type the rest of the line; other lines are the command output, you don't have to type them either.

Because all the configuration commands are important, you'll need to use a system administrator shell, like root, on the source and the target computers.

     $ su
     Password: blabla

The prompt will be shown as "$" in the remainder of this documentation "$", even if it should be "#". This is because "#" often means comment, so it is ambiguous. I don't like ambiguity in computer science.

8. Building your own cable

Extract from /usr/src/linux/Documentation/networking/PLIP.txt, written by Donald Becker (

Parallel Transfer Mode 0 Cable

The cable for the first transfer mode is a standard
printer "null" cable which transfers data four bits at a time using
data bit outputs of the first port (machine T) connected to the
status bit inputs of the second port (machine R).  There are five
status inputs, and they are used as four data inputs and a clock (data
strobe) input, arranged so that the data input bits appear as contiguous
bits with standard status register implementation.

A cable that implements this protocol is available commercially as a
"Null Printer" or "Turbo Laplink" cable.  It can be constructed with
two DB-25 male connectors symmetrically connected as follows:

    STROBE output       1*
    D0->ERROR   2 - 15          15 - 2
    D1->SLCT    3 - 13          13 - 3
    D2->PAPOUT  4 - 12          12 - 4
    D3->ACK     5 - 10          10 - 5
    D4->BUSY    6 - 11          11 - 6
    D5,D6,D7 are   7*, 8*, 9*
    AUTOFD output 14*
    INIT   output 16*
    SLCTIN      17 - 17
    extra grounds are 18*,19*,20*,21*,22*,23*,24*
    GROUND      25 - 25
* Do not connect these pins on either end

If the cable you are using has a metallic shield it should be
connected to the metallic DB-25 shell at one end only.

I checked my cable twice. On both sides, pins are connected like this:

     1 -  1 Yes
     2 - 15
     3 - 13
     4 - 12
     5 - 10
     6 - 11
     7 not connected
     8 not connected
     9 not connected
    10 -  5
    11 -  6
    12 -  4
    13 -  3
    14 - 14 Yes
    15 -  2
    16 - 16 Yes
    17 - 17
    18 not connected
    19 not connected
    20 not connected
    21 not connected
    22 not connected
    23 not connected
    25 - 25 not connected to metallic shield

Unlike the cable described in the kernel document, my cable has pins 1, 14, 16 connected. It doesn't matter, apparently, since plip works fine for me.

9. Network parameters

We will make a plip point-to-point network with this IP configuration:

You should not already have those names in your name space:

     $ ping source
     ping: unknown host source
     $ ping target
     ping: unknown host target

You should not already have those IP addresses in your network space:

     $ ping
     PING ( 56 data bytes
     ping: sendto: Network is unreachable
     ping: wrote 64 chars, ret=-1
     --- ping statistics ---
     1 packets transmitted, 0 packets received, 100% packet loss
     $ ping
     PING ( 56 data bytes
     ping: sendto: Network is unreachable
     ping: wrote 64 chars, ret=-1
     --- ping statistics ---
     1 packets transmitted, 0 packets received, 100% packet loss

If those commands don't give you errors, change the names or the addresses.

You can choose other addresses, names or netmask (netmask must be the same on both sides). In the rest of this document, I'll suppose you choose these addresses and these names.

Add a line in /etc/exports

     #### file /etc/exports ####
     /cdrom          source(ro) target(ro)
     #### EOF ####

Because the portmapper tries to resolve IP addresses, add the IP addresses and names in /etc/hosts

     #### file /etc/hosts ####
     ...     target     source
     #### EOF ####

Verify you have the item files for the hosts search list in the file /etc/nsswitch.conf

     #### file /etc/nsswitch.conf ####
     hosts:      files nis dns
     #### EOF ####

10. PLIP on the Source side

This section describes how to set up the plip interface in the source server. If you run into trouble, I suggest that you read the PLIP MINI-HOWTO.

Check that your lp device is not set. You should not have this entry:

     $ cat /proc/devices 
     Character devices:
     6 lp

If you do have it, kill the lpd daemon and remove the lp module:

     $ /etc/rc.d/init.d/lpd.init stop
     Shutting down lpd: lpd

     $ rmmod lp

If you can't remove the lp module then you have to recompile the kernel with lp service as a module.

Now, the "6 lp" line has disappeared from the /proc/devices file, which is a reflection of the kernel capabilities.

You are not obliged to eliminate the lp device : the scheme may work with lp. Without guarantee (it works for me). Check it yourself.

Check that your parallel port is handled:

     $ ls /proc/parport/
     $ cat /proc/parport/0/hardware 
     base:   0x378
     irq:    7
     dma:    none
     modes:  SPP,ECP,ECPEPP,ECPPS2

If you don't have any directory under /proc/parport/ then you have to load the parport and parport_pc modules:

     $ insmod parport
     $ insmod parport_pc

You should see this new entry in /var/log/messages:

     Oct  9 20:50:47 louloutte kernel:
     parport0: PC-style at 0x378 [SPP,ECP,ECPEPP,ECPPS2]

     Oct  9 20:50:47 louloutte kernel: 
     parport0: detected irq 7; 
     use procfs to enable interrupt-driven operation.

I repeat the message "detected irq 7, use procfs to enable interrupt-driven operation", so:

     $ echo 7 >  /proc/parport/0/irq

Using a kernel 2.4 the last command is no longer available. Use instead:

        $ insmod parport
        $ insmod parport_pc io=0x378 irq=7

Check that plip module is loaded:

     $ lsmod |grep plip

If plip module is not loaded, then load it:

     $ insmod plip

You should see something like this in /var/log/messages

     ==> /var/log/messages <==
     Oct  8 16:34:12 louloutte kernel:
     NET3 PLIP version 2.3-parport
     Oct  8 16:34:12 louloutte kernel: 
     plip0: Parallel port at 0x378, using IRQ 7

If you can't load the plip module then you have to recompile the kernel with plip service as a module.

The syslog message says the module is loaded on the plip0 interface. Configure the plip0 interface:

     $ ifconfig plip0 source pointopoint target netmask up

Check that everything is okay.

     $ ifconfig plip0
     plip0     Link encap:10Mbps Ethernet  HWaddr FC:FC:C0:A8:00:02
          inet addr:  P-t-P:  Mask:
          UP POINTOPOINT RUNNING NOARP  MTU:1500  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0
          TX packets:0 errors:0 dropped:0 overruns:0
          Interrupt:7 Base address:0x378 

Now you can ping locally the source server:

     $ ping source
     PING source ( 56 data bytes
     64 bytes from icmp_seq=0 ttl=255 time=0.3 ms
     64 bytes from icmp_seq=1 ttl=255 time=0.2 ms
     --- source ping statistics ---
     2 packets transmitted, 2 packets received, 0% packet loss
     round-trip min/avg/max = 0.2/0.2/0.3 ms

Verify that the route to target exists:

     $ route
     Kernel IP routing table
     Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
     target          *      UH    0      0        0 plip0

If the route doesn't exist, add it:

     $ route add -host  dev plip0

When the target is configured you will be able to do a ping test:

     $ ping target
     PING ( 56 data bytes
     64 bytes from icmp_seq=0 ttl=64 time=4.5 ms
     64 bytes from icmp_seq=1 ttl=64 time=4.3 ms
     --- ping statistics ---
     2 packets transmitted, 2 packets received, 0% packet loss
     round-trip min/avg/max = 4.3/4.4/4.5 ms

But if you try it now you should have:

     $ ping target
     PING target ( 56 data bytes
     --- target ping statistics ---
     5 packets transmitted, 0 packets received, 100% packet loss

Now, the server network is ready to work. Congratulations.

11. The cdrom nfs server side

This section describes how to mount and export via NFS a cdrom drive.

Mount the cdrom. If you encounter a problem with your cdrom drive, read the CDROM-HOWTO. I assume that the cdrom device is /dev/hdd but it could be /dev/sr0 or /dev/hdb etc. The mount point I choose is /cdrom but you can choose the one you want:

     $ mount -t iso9660 /dev/hdd  /cdrom

You have to set up nfs services. If something goes wrong, read the NFS-HOWTO

Verify your kernel supports nfs:

     $ cat /proc/filesystems 
     nodev   proc
     nodev   nfs
     nodev   ncpfs
     nodev   devpts

Verify your portmapper can handle mountd and nfs client requests:

     $ rpcinfo -p
     program vers proto   port
     100000    2   tcp    111  portmapper
     100000    2   udp    111  portmapper
     100005    1   udp    655  mountd
     100005    1   tcp    657  mountd
     100003    2   udp   2049  nfs
     100003    2   tcp   2049  nfs

Lines with nfs or portmapper have to be there. We have NFS on tcp? progress have been made!

It doesn't matter if you have only NFS on udp. NFS has to work, that's the thing you need.

Then, rerun portmap, mountd, nfs:

     $ /etc/rc.d/init.d/portmap.init stop
     Stopping INET services: portmap 

     $ /etc/rc.d/init.d/portmap.init start
     Starting portmapper: portmap

     $ /etc/rc.d/init.d/nfs stop
     Shutting down NFS services: rpc.mountd rpc.nfsd 

     $ /etc/rc.d/init.d/nfs start
     Starting NFS services: rpc.mountd rpc.nfsd 

/etc/rc.d/init.d/ is /sbin/init.d/ on SuSE Linux systems. I had a dream last night: LSB was respected. (LDB: Linux Standard Base)

Then, try to mount it "locally" (via nfs) on the source box and when the test is good remove the test directory:

     $ mkdir /tmp/nfstest
     $ mount -t nfs source:/cdrom /tmp/nfstest
     $ ls  /tmp/nfstest/
     README             debian/            locatedb.3         tools/
     TRANS.TBL  ls_lR.3            upgrade/
     boot/              install/           realaudio/
     $ echo great stuff !
     $ umount /tmp/nfstest
     $ rmdir /tmp/nfstest

Bad luck ? Read the NFS-HOWTO and /var/log/messages.

Check the nfs server with those commands:

     $ rpcinfo -p

Run portmap with the -v flag:

      $ portmap -v

kill portmap, mountd, nfsd and rerun them in this order: portmap, mountd, nfsd. Take a rest, get more coffee. Read again the NFS-HOWTO and restart.

Now, the exported cdrom drive nfs server is ready to work. Congratulations.

12. FIPS, PARTED, splitting of hard disk partitions

This section is for anybody who has only one primary FAT partition (DOS, Windows 3.xx, NT) and wants to keep it without loosing data. Make some backups because if you burn it, I won't be responsible - you will.

Journey to PARTED

Andrew Clausen (, GNU Parted maintainer, wrote "You might want to mention GNU Parted in the PLIP-Install Howto." That is done. Since I didn't try it I can only give you the links to parted by http and parted ftp. Help yourself. Parted seems very good.

Andrew said: "GNU Parted is a *LOT* more powerful than FIPS (think Partition Magic). For example, it can convert FAT16 <=> FAT32, change the size of the FATs, doesn't require a defragger, etc. And it supports ext2, linux-swap..."

Back to FIPS.

Download FIPS. Don't forget the mirrors where xx is your country abbreviation (fr, fi, us, uk, etc.).

At the time of this writing the archive file is called

The fips program is already in the debian cdrom distribution. I found it (yes, after downloading the file) in /cdrom/debian/tools/ (It supposes your cdrom is mounted on /cdrom)

I guess you are on a Unix world but you're not compelled to be. Go in a good working place on the source computer:

     $ mkdir /tmp/fips-2.0/
     $ cd /tmp/fips-2.0/
     $ unzip -l /archive/fips/
     $ unzip  /archive/fips/
     $ ls
     $ dos2unix fips.doc fips.faq  readme.1st 

  1. Read the file readme.1st
  2. Read the file fips.doc
  3. Read the file fips.faq

Boot your target computer. Be in pure DOS (quit Windows).

Read again the file fips.doc from the section "5. Before you start".

Be aware of hidden files.

Graham's comment: Hidden files will not be moved by DeFrag (as far as I know), so FIPS will reclaim less space for use by Linux. Use the ATTRIB command to remove the "hidden" attribute. Some things may require that files are hidden, so it will be necessary to hide the same files again later. I think that it would also be a good idea to run scandisk at this moment. Perhaps I do not trust DOS.

Be aware to eliminate the "virtual memory" file (swap for Win*) during the fips process. In Windows 3.11 (quite up to date, no?) this swap file is configured from Program-Manager->Control-Panel->Enhanced (a i386 chip icon).

Click, Click->Click on the chip and re-click on a button called "Virtual Memory".

Adjust the size to none, Click<-Click<-Click<-Click back plus Alt-F4 to close all your windows. Don't you think the Gates are closed too?

Graham's comment: I am not sure that this is a good idea. What happens if "386spart.par" is not hidden? I think that DEFRAG will move it, and Windows will complain next time it starts, then rebuild the file. Since we are clearly keeping DOS/Windows, the user will require a swap file again later. By not deleting it, the disc space is reserved for that time.

Graham's new comment: "Make sure that the file is not hidden (attrib -h -s 386spart.par), then delete it (del 386spart.par). The disc space is then free. Run FIPS. Next time Windows starts, it will complain that the file is corrupt, and will recreate it."

I don't what to say. I've just parroted FIPS manual because I succeeded with it.

So drink coffee, and pray.


     C:\> SCANDISK

Correct every cluster. You should have no dead cluster now.


Run DEFRAG on C:

     C:\> DEFRAG C:

All used clusters are at the beginning of the hard disk.


Make a bootable floppy disk:

     C:\> FORMAT A: /S

Copy AUTOEXEC.BAT and CONFIG.SYS to the new floppy if you want to keep your local features (keyboard keys etc.). Remove the line running SMARTDRIVE.EXE in AUTOEXEC.BAT. Keep a minimal AUTOEXEC.BAT

On the source box: Copy the fips files restorrb.exe, fips.exe and errors.txt to this floppy disk.

    $ mcopy errors.txt fips.exe restorrb.exe a:

Graham's comment: Doesn't this assume that mtools are installed? If the floppy disc is mounted as type msdos, the standard "cp" command should do the job, and mtools would not be needed. I have never used mtools, but I have played with DOS files from Linux.

Yes, this assumes that mtools are installed. mcopy is faster to type than mkdir, mount, cp, umount, rmdir.

Boot the target with the floppy. Run fips while reading the section "Using FIPS" in the file fips.doc

Do not forget to answer yes when fips asks for a rootboot saving.

When everything is ok, save your hard disk space:

     $ cd
     $ rm -rf /tmp/fips-2.0/

Now you have a nice free space for a new operating system.

13. Debian installation

I'm a Debian proselyte since it is my first Debian installation. Debian lovers are the same kind of people as Macintosh lovers or Linux lovers in the operating systems space. Nothing else counts. Since I'm already a Mac and Linux lover (and French too :-) it was time to fall in love.

Debian is well, clearly and internationally documented. Thanks to all those people who bred this really open distribution.

I could leave you on your own during the installation process. But since we have to interact with a shell during it, the entire process will be described in detail.

I know the description is Debian specific. I prefer to give you a complete example than nothing except a ``run a shell at the right moment and type bla bla...''. I do like concrete examples.

Simon Forget <> told me he could not use this howto with his Toshiba Libretto 50CT because the kernel could not recognize the pcmcia floppy drive during the installation process. I don't know why but there is a solution using plip, if you already have a dos partition on the target computer.

This solution is simpler and faster because no floppy drive nor plip network is necessary during the installation process. If you are interested in this solution, go directly to the section Installing from a DOS partition. I'm keeping the old installation routine because this new one needs a DOS partition that becomes superfluous when you definitely want to leave the dark side.

13.1 Preparing the two floppies

On the source side, mount the cdrom and go to the install directory

      $ mount -t iso9660 /dev/hdd  /cdrom

The device for your CDROM may be /dev/hdb or /dev/hdc or another device (SCSI), depending on where is plugged your CDROM.

      $ cd /cdrom/debian/dists/stable/main/disks-i386/current/

Read the file install.html with a browser or install.txt with a more, a less, a cat (or a dog?).

Now, record the install (alias rescue) floppy. Write a "resc1440.bin" label on it:

      $ dd if=resc1440.bin of=/dev/fd0H1440

Write the drivers floppy. Write a "drv1440.bin" label on it:

      $ dd if=drv1440.bin of=/dev/fd0H1440

Then make sure that the floppies are write-protected.

Now you're ready to start the real installation process.

13.2 The real installation process

Insert the resc1440.bin floppy on target drive. Reboot your target box.

     Welcome to Debian GNU/Linux 2.x!

Read the text. The prompt is:


Press <ENTER>

     Loading root.bin...........
     loading linux...

A new screen:

     Next: Select Color or Monochrome display

Choose yours with the arrow up/down keys and press <ENTER>

     Next: Continue with the installation

press <ENTER>

     Software in the Public Interest 
      *** Debian GNU/Linux 2.1 ***

Read the text Press <ENTER> (You see <Continue> on the screen)

     Next: Configure the Keyboard

press <ENTER> Select your country and press <ENTER>

     Next : Partition a Hard Disk

Press <ENTER>

     Select Disk Drive 
     usually /dev/hda

If you have several disks choose the good one (where you want to install Linux) and press <ENTER>

You are now using the program cfdisk. Using cfdisk is safe until you decide to [Write] the partition on disk. Leaving cfdisk with [Quit] is safe.

We are going to make just two new primary partitions, a Swap one and a big Linux one. If you want to create more partitions, leave some free space or extended partitions, you can. But read the documentation about "Partitioning" in the Installation-HOWTO, especially if you have more than 1024 cylinders and an old version of LILO.

If you previously ran the FIPS program, you see the hda1 partition (DOSFAT16). Switch to the next free partition with the up/down arrow keys.

Select [New] with the right/left arrow keys and press <ENTER>

Select [Primary] and press <ENTER>

Enter the size of your swap partition. Twice the RAM is usual if you have less than 128 megabytes of ram. If you have 2 gigabytes of RAM, it's because you don't want to swap. In that case, no swap partition is needed.

Select [Beginning] and press <ENTER>

Select [Type] and press <ENTER>. Type 82 (Linux Swap) and press <ENTER>.

Switch to the next free partition with the up/down arrow keys. Select [New] and press <ENTER>

Select [Primary] and press <ENTER>

Enter the size in MB (you can leave the default) and press <ENTER> Its type should be already Linux. If not, change to Linux (83) with [Type]

The partition table is defined now. Verify everything looks good. If you're not sure, read the documentation with [Help]. If doubts are still there, select [Quit] and leave the installation process. Go for a walk and restart from the beginning of this section.

I assume you are confident now.

Select [Write]

     Are you sure you want to write the partition table to disk?

Type "yes" and press <ENTER>

Select [Quit] and press <ENTER>

     Next: Initialize and Activate a Swap Partition

Press <ENTER>

     Please select the partition to initialize as a swap device

Select /dev/hda2 (normally already selected). Press <ENTER>

     Scan for Bad Blocks?

select <YES> and press <ENTER>

     Are you Sure?

select <YES> and press <ENTER>

     Initializing swap partition

     Next: Initialize a Linux Partition

Press <ENTER>

     Select Partition. Please select the partition to initialize as a
     Linux "ext2" file-system.

Select /dev/hda3 (normally already selected) and press <ENTER>.

     Scan for Bad Blocks?

Select <YES> and press <ENTER>

     Are you Sure?

Select <YES> and press <ENTER>

A new page full of numbers. You can take a rest because it takes some time (especially with a big hard disk).

     Next: Mount a Previously-Initialized Partition

Press <ENTER>

     Please select the partition to mount

Select /dev/hda3 and press <ENTER>

     Mount the /dev/hda3 device as the Root FileSystem?

Select <Yes< and press <ENTER>

     Next: Install Operating System Kernel and Modules

Press <ENTER>

     Please select the medium you will use to install the system

Select /dev/fd0 and press <ENTER>

     Please place the Rescue Floppy in the first floppy drive

The floppy is already there.
Select <Continue> and press <ENTER>

     Installing the Rescue Floppy ...
     Please place the Drivers Floppy in the first floppy drive

Eject the Rescue Floppy and insert the Drivers Floppy, the one I you labelled drv1440.bin (you did it, didn't you?).
Select <Continue> and press <ENTER>

     Installing the Drivers  Floppy ...

A new screen:

     Next: Configure Device Driver Modules

Press <ENTER>

     Select Category

Read the text.

     Please select the category of modules

Select net and press <ENTER>
Select plip.

     Module plip

Select "Install the module in the kernel" and press <ENTER>
No parameters are needed.
Select <Ok> and press <ENTER>

     Installation succeeded
     Please press ENTER when you are ready to continue.

Press <ENTER>

Select Exit ("Finished with these modules") and press <ENTER>
Then you see the same kind of screen again.
Select Exit ("Finished with these modules") and press <ENTER>

     Next: Configure the Network

Choose a name, you can use a different name than debian or target.
Press <ENTER>

     Is your system connect to a network? 

Select <No> and press <ENTER>

STOP STOP STOP. You see now:

     Next: Install the base system

Now we need a shell.

Press Alt F2 and <ENTER> You are in a root shell.

13.3 An installation break: PLIP on the target side

You are in a root shell.

Verify the plip module is loaded:

     $ lsmod
     Module    Pages    Used by
     plip          3          0

Find the exact name of the plip interface:

     $ dmesg
     NET3 PLIP version 2.2
     plip1: Parallel port at 0x378, using assigned IRQ 7

You can find the same information with

     $ cat /proc/kmsg
     <4>NET3 PLIP version 2.2
     <4>plip1: Parallel port at 0x378, using assigned IRQ 7
     Ctrl-c (or ^C if you prefer)

Configure the plip interface:

     $ ifconfig plip1  pointopoint netmask up

Verify everything is ok:

     $ ifconfig plip1 
     plip0     Link encap:Ethernet  HWaddr FC:FC:C0:A8:00:01
          inet addr:  P-t-P:  Mask:
          UP POINTOPOINT RUNNING NOARP  MTU:1500  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
          Interrupt:7 Base address:0x378

Verify the route to source exists:

     $ route 
     Kernel IP routing table
     Destination     Gateway         Genmask         Flags Metric Ref    Use Iface       U     0      0        0 lo

If not, like in the previous example, add the route to, the source:

     $ route add -host dev plip1

Now the route is installed:

     $ route 
     Kernel IP routing table
     Destination     Gateway         Genmask         Flags Metric Ref    Use Iface UH    0      0        0 plip1       U     0      0        0 lo

Now you can perform a successful ping from the source server (ping is not available on the Debian install process):

     $ ping target
     PING target ( 56 data bytes
     64 bytes from icmp_seq=0 ttl=64 time=14.0 ms
     64 bytes from icmp_seq=1 ttl=64 time=4.3 ms
     --- target ping statistics ---
     2 packets transmitted, 2 packets received, 0% packet loss
     round-trip min/avg/max = 4.3/9.1/14.0 ms

If it doesn't work, check the Null-Modem cable connection, rerun dmesg, ifconfig, route. Verify everything. Take a rest. Restart.

When the ping is okay, return to the normal installation process:

On the target: press Alt F1

13.4 Return to the normal install process

You are back to the normal installation process. You should still see:

     Next: Install the base system

Press <ENTER>

     Please select the medium you will use to install the system

Select nfs and press <ENTER>

     Please choose the NFS server and the mount path ...

Press <ENTER>

     Please choose the path inside the mounted NFS filesystem

Type /debian (normally already there) and press <ENTER>

If no error message complains about the nfs mount then bravo. You can verify this mount by entering the shell again:

Press Alt F2

     $ mount
     ... on /instmnt type nfs (rw, addr=

     $ ls /instmnt/debian
     README                 README.non-US          doc/
     README.CD-manufacture  README.pgp@            hamm/
     README.mirrors.html    TRANS.TBL              tools/
     README.mirrors.txt     dists/

Press Alt F1

You're back again to the normal installation process.

Now the plip nfs cdrom connection is done. Let's go on and finish our job.

     Please select the directory containing a file base2_0.tgz

Select list and press <ENTER>

     Please Wait 
     The installation program is building a list of ...

A new screen:

     Select Archive Path
     Please select the directory that you will use to install the Base
     System from.

Only one long item, already selected. Press <ENTER>

     The Base System is being extracted from

Take a second rest, you deserve it

     Next: Configure the Base System

Press <ENTER>

     Select Timezone

Select your timezone and your directory and press <ENTER>

     Timezone Configuration

Read and press <ENTER>

     Timezone Configuration.

An other page. Read, select <YES> and press <ENTER>

     Next: Make Linux Bootable Directly From Hard Disk

Press <ENTER>

     Create Master Boot Record?

Select <Yes> and press <ENTER>

     Make Linux the Default Boot Partition?

Select <No> and press <ENTER>

     Next: Make a boot Floppy

Press <ENTER>

     Change Disk
     Please place a blank floppy disk in the first floppy drive.

Do it and press <ENTER>
The floppy is being formatted

     Creating a filesystem on the floppy...
     Copying the operating system kernel...

A new screen:

     Next: Reboot the System

Press <ENTER>

     Reboot the system ?

Remove the floppy and press <ENTER>

The system reboots. Are you still with the dark side? I guess yes, so insert the boot floppy you've just made and reboot again with Ctrl-Alt-Del

See the boot messages.
Read the text.

     New password:

Enter a root password.

     Re-enter new password

Do it. And learn it.

     Shall I create a normal user account now? [Y/n]

Enter n and press <ENTER>. Unless you want to do it (it is safe and good, in fact).

     Shall I install shadow passwords? [Y/n]

Enter y and press <ENTER>

     Do you want to use a PPP connection to install

Enter n and press <ENTER>

     Now you may choose one of several selections ...
     Do you want to perform this step?

Enter n and press <ENTER>

     I'm going to start the 'dselect' program...

Press <ENTER>
Select [Q]uit and press <ENTER>

     You may now login as 'root' at the login: prompt...
     debian login:

Enter root.


Enter the root password.


The system is installed and working. CONGRATULATIONS!

My job stops here. Read the install documentation of your distribution and go on with the Unix system administration job.

When you reboot your system, the plip connection won't be in good shape. But now you know what to do.

I suggest you some work:

14. Installing Slackware, Caldera

This section is written by Florent SAUNIER.

Here, '&' is used for the root prompt, rather than '$'.

I have performed a Slackware installation using PLIP very succesfully. I have also been able to install RedHat and Caldera Open linux using the very same method described below.

As previously specified by Gilles, a Slackware installation requires everything specified in the chapter "What do you need?" with the time increased to 10 hours if you desire to install xfree(86) and some development tools.

Further on, you first need to read Gilles' installation as I set up mine according to his method.

Finally, I have performed Slackware and Caldera installation on a Twinhead 550C laptop, with 8Mb Ram, and 2Gbytes Hard drive.

14.1 Slakware installation

Making boot,root and plip disks

To install Slackware verion using an ATAPI CDROM from your source, you will need the standard kernel image bareapm.i or bare.i. If you want to make an install with SCSI support, use the scsinet.i image.

Other images exist, such as net.i which should include the PLIP function but on my i486 laptop, Twinhead550C, the image does not work. In my experience, the standard image always works on any laptop I have used.

1a- creating the boot and root disks.

From DOS (PC-DOS, DR-DOS, FreeDOS, ...) create the boot disk by typing:

    rawrite bareapm.i
then create the root disks:
    rawrite color.gz.a  (if you are using a color screen)

1b- Creating the PLIP disk

On a DOS formated disk copy the following files:

Those modules are compulsory prior to a PLIP install, you will find them into the Slackware installation CDROM under the directory /MODULES. The same applies to the Caldera installation. Run a "find" if you cannot find them.

1c- launching the installation.

Insert your boot disk in your Laptop and start the computer. Follow the installation procedure and wait until the system ask for the root disk. Insert it and wait for the prompt login: "Slackware login".

To start the Slackware installation you must log in as root. Type "root" at the prompt. Be aware that at the time of installation your keyboard is still qwerty.

2a- Install PLIP module

Install your source computer according to Gilles' instructions, from chapter 7 to 9.

Install PLIP on the target. To install the PLIP module you need to launch the program from your PLIP disk. Remember you did it, no !!!

Access your floppy disk with:

          $ mount /dev/fd0h1440 /floppy
          $ cd /floppy
          $ insmod parport      # it will install your parallel port
          $ insmod parport_pc   # it will set it up
          $ insmod plip         # it will install plip module
If during the PLIP installation you encounter an error, it means that your parallel port has not been correctly installed. You may type again the instruction above, it does work sometimes !!, or go to the parport directory and write the appropriate setting.
      $ cd /proc/parport/0
      $ vi IRQ            # Verify that 7 is written in it if you are using
                          # that interrupt'
See the chapter "PLIP on the source side" to get the proper setting.

You should not bother too much as at 95% on the 5 laptops I have installed everyting went smoothly. (Comment peut-on avoir "95% de cinq", said rational Graham)

3a- Now create your NFS drive.

From that point I have chosen to use the creation of a NFS drive on the target computer. Then I will make an installation using the "From directory" option rather than from nfs.

Do not forget you still have a US keyboard layout. Of course you have set up your source according to chapter "Network parameters". You did then ping your source, but you cannot yet ping the target.

So the first step is to configure your PLIP connection: On the target type:

       $ ifconfig plip0 pointopoint netmask up

Test your Plip configuration by pinging the target from the source

     $ ping target
then mount your nfs drive on the target:
     $ cd /
     $ mkdir nfstest
     $ mount -t nfs /nfstest
You have just configured a drive linked to your source directory /cdrom were you suppose to have the Slackware CDROM connected, from your target directory called nfstest.

Now everything is set up and you are ready to launch the Slackware installation program, by typing "setup".

note:if you do not have enough memory, below 8MB setup will not be launched, then you will have to make a swap space to allow the program to work. First, create your disk partition with your swap by typing

       $ fdisk

Note which is your swap partition (e.g. /hda3), then quit fdisk with 'w'. Once it's done, type:

       $ mkswap -c /dev/hda3
       $ swapon /dev/ha3

Your swap is now set up. You may then launch the "setup" program.

4a- Installing Slakware

the setup programm shows various menu items:

4b- Select your keyboard, for French take

4c- Select your swap partition if not already configured

4d- Select the target, default is '/'

4e- Select the source. tricky stuff: choose "Installing from a drive", then type the drive so : /nfstest/slackware '/slackware' because your drive must point to your slackware CDROM directory

4f- Select packages to install

4g- Wait, Wait, Wait and Wait........

14.2 Caldera Installation

I told you that Caldera works also with that install. In fact, as Caldera automatically launches its Setup program, once you arrived at the menu where you are supposed to choose your source disks, type ALT F2, go to another screen, then log on as root and make your nfs drive as described above in points 2 to 4. Then go back to the installation screen with ALT F1, and type the famous nfstest as your source directory. Caldera will then install without problem.

14.3 Troubleshooting

If you are experiencing any trouble with that installation you may report it to me and I'll try to help as best as I can.

15. Installing RedHat 6.x

This section is written by Dr. Tilmann Bubeck,

I tried hard to install a laptop through PLIP with a RedHat 6.2 but the installer from RedHat is buggy and could not be used for PLIP. I verified that it was already broken in RedHat 6.1, but works in RedHat 5.2 as desired. Patches have been sent to RedHat and will hopefully be applied to the upcoming RedHat 7.0.

So if you want to install a RedHat 6.x over PLIP you must follow these instructions.

The problem is the "bootnet" disk. This disk should be used for PLIP installation but has 2 serious errors:

15.1 Using a fixed "bootnet" disk

Probably the easiest way to solve the above problem is to use a fixed "bootnet" disk. You can get a fixed "bootnet" disk for RedHat 6.2 from

Use the following command to copy the downloaded image to an empty disk:

dd if=bootnet-plip-62.img of=/dev/fd0.

Then insert the disk and turn your computer on. When you come to the point where you must choose your network device driver please do:

15.2 Building your own "bootnet" disk

If the above bootnet disk is not suitable for your needs, I will present a small description of how to transform a standard bootnet disk for using "plip". This description is quite short and is addressed to the advanced user.

By following these instructions, you get a bootnet disk containing all necessary modules and therefore allowing for a PLIP installation. Please follow the instructions elsewhere in this how-to on how to do the installation with the new disk.

16. Installing RedHat 7.0

This section is written by Leonard den Ottolander

To install RedHat 7.0 over PLIP the easiest way to go is to use a patched bootnet.img which is available via the url This saves the need for a driver disk, which might be useful when installing on a machine which floppy drive stops functioning after boot (I made this bootnet-plip.img to do an install to a Toshiba Libretto 30 which has a funny PCMCIA floppy drive that is no longer available after boot).

This disk is a full bootnet.img for RedHat 7.0, with the modules necessary for PLIP added to it. Since there was enough disk space, no modules had to be deleted. So you can use this image as if it were a full featured bootnet.img. No need for different disks.

I am not going into detail on the actual installation procedure, but there are a few things to note:

16.1 Creating your own bootnet-plip disk

This section is a slightly adapted copy of the procedure to create a bootnet-plip image for RedHat 6.2 described by Tilmann Bubeck. Thanx Tilmann!

17. Installing Suse

The Suse installation process includes PLIP support (Network type configuration). If you encounter a problem installing a Suse distibution using PLIP, feel free to report them.

18. Installing from a DOS partition

You can use this section if you already have a DOS partition on your target computer. I think it is a faster installation method.

Read the section 5.3.1 "Installing from a DOS partition" from install.txt. This section is also available on the Debian CD in the html file ch-install-methods.html.

I am just going to help you to make the first point : "1. Get the following files from your nearest Debian repository"

The transfer is easy with the floppy-disc based Linux from  tomsrtbt.

Export the source cdrom drive via nfs.

Read the file tomsrtbt.FAQ

Create the floppy under DOS or Linux, you have the choice.

     Linux installation:
     a) extract the .tar.gz archive
     b) Be root
     c) Be in the tomsrtbt-<version> directory
     d) Have a blank floppy with no bad sectors
     e) Do './install.s'

Boot the target with the floppy. Login as root.

      $ insmod plip
      $ ifconfig plip1 pointopoint \
      netmask up
      $ route add -host dev plip1
      $ mount -t nfs /cdrom
      $ mkdir /c
      $ mount -t msdos /dev/hda1 /c
      $ mkdir /c/debian
      $ cd /cdrom/dists/stable/main/disks-i386/current/
      $ cp resc1440.bin drv1440.bin base2_1.tgz root.bin linux \
      install.bat loadlin.exe /c/debian

The \ means the command line continues on the next line. The cp command takes some time. When finished, remove the floppy and then :

      $ reboot

Boot under DOS.

      C:\> cd debian
      C:\DEBIAN> install

Now you start to install Debian from a DOS partition. You do not need plip or floppy during the installation process. Everything deals with the hard drive. Read the Debian install documentation, it is very good and clear.

19. Install the plip interface permanently

19.1 On the source side

I use an old Linux RedHat 4.1 distribution. The location of the files can be different on other GNU/Linux distributions but the philosophy is the same (The Unix System V convention).

Create the file /etc/rc.d/init.d/plip with this content:


# file /etc/rc.d/init.d/plip #

# See how we were called.
case "$1" in
        # Start daemons.
        /bin/echo "Starting plip interface: "
        /bin/echo "Doing /sbin/ifconfig plip0 source pointopoint target netmask up"
        /sbin/ifconfig plip0 source pointopoint target netmask up
        /bin/echo  "Doing /bin/ping -q -c 4 target"
        /bin/ping -q -c 4 target
        /bin/echo "Starting plip interface: done"
        # Stop daemons.
        /bin/echo  "Shutting down plip interface:"
        /bin/echo  "Doing /sbin/ifconfig plip0 source pointopoint target netmask down"
        /sbin/ifconfig plip0 source pointopoint target netmask down
        /bin/echo  "Doing /sbin/modprobe  -r plip "
        /sbin/modprobe  -r plip
        /bin/echo "Shutting down plip interface: done"
        echo "Usage: $0 {start|stop}"
        exit 1

exit 0

# === End of File ===


Only the ifconfig lines are strictly necessary. Perhaps you will need to add some modprobe commands if you don't use kerneld or the kmod feature of new kernels 2.2.x

Create the symbolic links in the rc*.d directories:

      $ cd /etc/rc.d/rc0.d/
      $ ln -s ../init.d/plip K97plip
      $ cd /etc/rc.d/rc1.d/
      $ ln -s ../init.d/plip K92plip
      $ cd /etc/rc.d/rc3.d/
      $ ln -s ../init.d/plip S11plip

      $ cd /etc/rc.d/rc5.d/
      $ ln -s ../init.d/plip S11plip

You can choose other numbers. Make sure that the two-digit number after 'K' is greater than the number of every other file that stops a service depending on plip.

Make sure that the two-digit number after 'S' is less than the number of every other file that start a service depending on plip: nfs, nis, ftp, http etc.

Update the /etc/conf.modules file, choosing the correct IRQ number (7 is mine, yours may be different):

# /etc/conf.modules
alias parport_lowlevel parport_pc
post-install parport_pc echo 7 >  /proc/parport/0/irq

Test the plip shell:

      $ /etc/rc.d/init.d/plip 
      Usage: /etc/rc.d/init.d/plip {start|stop}

      $ /etc/rc.d/init.d/plip stop
      Shutting down plip interface: 
      Doing /sbin/ifconfig plip0 source pointopoint target netmask down 
      Doing /sbin/modprobe  -r plip 
      Shutting down plip interface: done

      $ /etc/rc.d/init.d/plip start
      Starting plip interface: 
      Doing /sbin/ifconfig plip0 source pointopoint target netmask up 
      Doing /bin/ping -q -c 4 target
      PING target ( 56 data bytes
      --- target ping statistics ---
      4 packets transmitted, 4 packets received, 0% packet loss
      round-trip min/avg/max = 4.4/8.3/14.0 ms
      Starting plip interface: done

Updating the start scripts is a good occasion to reboot a Unix system, to check the modifications. Do it:

      $ init 6 # or "shutdown -r now" or "reboot"

19.2 On the target side

Update the file /etc/init.d/network:

      #! /bin/sh
      # /etc/init.d/network #

      ifconfig lo
      route add -net
      ifconfig plip1 pointopoint netmask up
      route add -host dev plip1

That's all because the parport features are directly in the kernel.

Updating the start scripts is a good occasion to reboot a Unix system, to check the modifications. Do it:

      $ init 6

20. Compatibility with a printer (lp)

I use the same parallel port for printing on a printer and networking with my laptop without big problems. It takes less than one second to switch from printing to networking. I just add a cheap manual switch between the parallel port, the printer and the laptop. The cable between the switch and the desktop computer is a fully connected one. The switch is just the one used to shared printers when no network is available. It works.

Sometimes, the kernel is not very happy because it can't finish its network job. In that case, I shut down the plip interface and restart it when needed.

I bought a new parallel card last year but it is still in its box. I'll write an "install a second parallel port" when I succeed in opening the box and add that beautiful card in the last connector I have left.

21. Need to compile a new kernel ?

Several times during the source server configuration we mentioned that a kernel compilation on the source computer could be needed.

I list the things you need to change or set. If you have never compiled a kernel, read the Kernel-HOWTO.

    Loadable module support  --->
        [*] Enable loadable module support

    General setup  --->
        [*] Networking support
        <M> Parallel port support
        <M>    PC-style hardware

    Networking options  --->
        [*] TCP/IP networking
    Network device support  --->
        [*] Network device support
        <*> Dummy net driver support
        <M> PLIP (parallel port) support
        <*> or <M>  PPP (point-to-point) support 

    Character devices  --->
        <M> Parallel printer support

    Filesystems  --->
        [*] /proc filesystem support
        <*> or <M> ISO 9660 CDROM filesystem support
        Network File Systems  --->
            <*> or <M> NFS filesystem support

22. Known problems

22.1 plip 2.2.14 versus 2.2.16

>From "Matt Haycraft"

Both sides have the right thing show up with ifconfig and route but ping (or anything else does not work) However what blows my mind is that the ifconfig section for both machines DOES show that packets are being received/transmitted.

However it turned out to be a compatibility issue between 2.2.14 and 2.2.16 and the plip code for each kernel version. 2.2.16 had some differences. To address the issue, I copied the plip.c from the 2.2.16 kernel to the source tree of 2.2.14 and recompiled the kernel module. Then everything worked!

23. What's new

This tag is automatic so it IS the reference. (Sometimes I forget to update the version number and the date)

$Id: PLIP-Install-HOWTO.sgml,v 1.40 2002/07/07 23:46:32 gilles Exp $

$Log: PLIP-Install-HOWTO.sgml,v $
Revision 1.40  2002/07/07 23:46:32  gilles
Corrected tag problem

Revision 1.39  2002/07/07 23:42:27  gilles
Added Richard Barrington in "acknowledgements" section.

Revision 1.38  2002/07/07 23:15:53  gilles
Added a "Known problems" section
Added "plip 2.2.14 versus 2.2.16" section. Thanks Matt.

Revision 1.37  2002/07/07 22:56:44  gilles
Adapted insmod parport_pc to 2.4 kernel

v1.36, 3 December 2001

v1.31, 21 April 2001

v1.30, 13 April 2001

v1.28, 18 December 2000

v1.27, 11 August 2000

v1.22, 16 June 2000

v1.15, 26 November 1999

24. To do

25. List of contributors and acknowledgements

I thank all the people in this list of contributors or projects. If you think I forgot someone, do not hesitate to complain.

Werner Heuser <>

He said: "yesterday I had time enough to read your document. It's great and in the next issue of my HOWTO (probably 2. December) I will make a link to it."

He did.

LDP <>

The Linux Documentation Project accepted the PLIP-Install-HOWTO without any hesitation.

Graham Bosworth <>

Graham corrected my bad English all over the document. Several times. Graham told me that Suse includes a PLIP installation process.

Simon Forget <>

Simon had a problem with his pcmcia floppy drive. So he led me search for a solution that is, in fact, simpler and faster (the DOS one).

Andrew Clausen <>

Andrew suggested that I talk about parted

Florent SAUNIER <>

Florent wrote the section called "Installing Slackware/Caldera".

Tomas Pospisek <>

Tomas also corrected my bad english.

Brad Rigby <>

Brad corrected a bug in the Slackware section (inverted IP adresses).

Donald Becker <>

Donald wrote the PLIP kernel documentation

Tilmann Bubeck <>

Tilmann wrote the section called "Installing RedHat 6.x".

Leonard den Ottolander <>

Leonard wrote the section called "Installing RedHat 7.x".

Richard Barrington (

"Under linux 2.4, the described method of setting irq doesn't work. it needs insmod parport_pc io=0x378 irq=7 otherwise it will default to no irq." Thanks Richard.

Matt Haycraft (

Matt discovered and resolved a plip problem between 2.2.14 and 2.2.16 kernels. See Known problems section.