OK - gory details. First, I have read the latest net-2-howto, so please
don't ask! Secondly, I understand it (so please don't ask :-). Thirdly,
I have a null modem cable (although I haven't opened it - it is one of
those ultra high-tech sealed jobs). Fourthly I can cat ascii to and fro
between the two computers along the cable. Fifthly, I know about setserial.
Sixthly there are no gettys running on the lines. Seventhly, yes, I
do have slip.o etc. loaded as kernel modules. Eighthly, I have read
Documentation/Changes.
At this very minute I am running identical (2.0.0) kernels on the two
machines for the purpose of comparison. I can step one of them up to
2.0.25 but I'll have to work on the other before I can do that to it.
======= same kernel: proof ===============
piano: uname -a
Linux oboe 2.0.0 #67- Mon Jun 10 15:06:15 MET DST 1996 i586
oboe: uname -a
Linux oboe 2.0.0 #67- Mon Jun 10 15:06:15 MET DST 1996 i586
The two machines are connected via their second serial ports. The uarts
are set to normal speed. One uart is a 450 (on the old 486). The other
is a 550 (the newer pentium).
======= uarts ===============
piano: setserial /dev/cua1
/dev/cua1, UART: 16450, Port: 0x02f8, IRQ: 3
oboe: setserial /dev/cua1
/dev/cua1, UART: 16550A, Port: 0x02f8, IRQ: 3
Both machines have slip.o loaded as modules. I did it with modprobe so
that they wouldn't be autocleaned away. One machine is mounting a dmsdos
filesystem over vfat, the other is not. One machine has a 3com ethernet
card, the other has an older card.
====== modules ===============
piano: lsmod
Module: #pages: Used by:
slip 2 0
slhc 2 [slip] 0
wd 2 1 (autoclean)
8390 2 [wd] 0 (autoclean)
nfs 12 8 (autoclean)
umsdos 5 1 (autoclean)
dmsdos 16 [umsdos] 1 (autoclean)
msdos 2 [dmsdos] 1 (autoclean)
vfat 3 [dmsdos] 0 (autoclean)
fat 6 [umsdos dmsdos msdos vfat] 1 (autoclean)
oboe: lsmod
Module: #pages: Used by:
serial 7 1 (autoclean)
slip 2 0
slhc 2 [slip] 0
nfs 12 6 (autoclean)
3c509 2 1 (autoclean)
umsdos 5 1 (autoclean)
Take it from me that there are no gettys on the lines, I am now going
to killall slattachs take the sl0 line down if it is there, then run
slattach and ifconfig. I am going to use the private network
192.168.1.0 for the interface. The older machine (piano) has
an older nettools and the slattach version is 1.1.79 (of 1995) whilst
the newer machine (oboe) has slattach 1.2 and the nettools that comes
with slackware 3.1. This is from strings ...
piano: NET-3 Base Utilities release tools-1.1.79
@(#) slattach 1.1.79 (10-Jan-95)
oboe: $Id: slattach.c,v 1.2 1996/07/15 23:06:34 dholland Exp $
Both machines run libc 5.4.7, ld.so 1.8.1
========== slattach attempt on piano ================
piano: slattach -s 38400 /dev/cua1 &
piano: ifconfig sl0 192.168.1.102 pointopoint 192.168.1.101
piano: route -add 192.168.1.101 dev sl0
This apparently succeeds.
piano: ifconfig
sl0 Link encap:VJ Serial Line IP
inet addr:192.168.1.102 P-t-P:192.168.1.101 Mask:255.255.255.0
UP POINTOPOINT RUNNING MTU:296 Metric:1
RX packets:0 errors:0 dropped:0 compressed:0
TX packets:0 errors:0 dropped:0 compressed:0
To make things less dubious, I'll try the older (same as on piano) slattach
first on oboe, then the newer one.
========== slattach attempt on oboe ================
oboe: slattach -s 38400 -p slip /dev/cua1
oboe: ifconfig sl0 192.168.1.101 pointopoint 192.168.1.102
oboe: route -add 192.168.1.102 dev sl0
and this apparently succeeds
oboe: ifconfig
sl0 Link encap:Serial Line IP
inet addr:192.168.1.101 P-t-P:192.168.1.102 Mask:255.255.255.0
UP POINTOPOINT RUNNING MTU:296 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0
TX packets:0 errors:0 dropped:0 overruns:0
and the routing table looks fine to me (if anyone is interested). I'll
edit the table to show only the sl0 entry on oboe. Piano is symmetrical.
Kernel routing table
Destination Gateway Genmask Flags MSS Window Use Iface
slippiano * 255.255.255.255 UH 296 0 0 sl0
loopback * 255.0.0.0 U 3584 0 9 lo
BUT I CAN'T PING/TELNET/FTP ANY FLAMING ONE OF 'em (sorry :-). Tcpdump
just sits there listening without any output.
OK - so now I'll take down slattach on oboe, kill sl0 and reinstall the
modules, then try the newer slattach. No. Sorry. Same result. The sl0
line comes up in ifconfig and then I can't do anything with it.
There are no error messages in syslog. I get an error message if I
restart slattach without having removed and reinstalled the slip.o
module first. That is all. It is:
Nov 24 19:17:33 oboe kernel: ERROR 'sl0' busy and not MOD_IN_USE.
Any clues? Other clues I have are that I cannot get ANY agetty/uugetty
to watch the lines for logins successfully. They just sit there while I
type characters at them. Two minicoms on the lines just about communicate
with each other.
I suspect that it has been such a long time since people used serial
cables between two computers that something subtle may be going wrong
nowadays. I retain the impression that I had serial connnections
working perfectly about kernel version 1.3.90 with slackware 3.0.
Peter T. Breuer
,---------------------------------------------------------------------------
|Departamento de Ingenieria de Sistemas Telematicos, Universidad Politecnica
|de Madrid, Escuela Tecnica Superior de Ingenieros de Telecomunicacion,
|Ciudad Universitaria, E--28040 Madrid, SPAIN.
|Tel. Office : +34 (1)336 6831
| Fax : +34 (1)543 2077 or 336 7333
|Internet : <ptb@eng.cam.ac.uk, ptb@comlab.ox.ac.uk, ptb@dit.upm.es>
| URL : http://www.dit.upm.es:80/~ptb/
`---------------------------------------------------------------------------