I have the pre-0.96 version of LINUX (0.95c++) running on a 20Mhz Gateway
2000 80386sx with a 110 Mbyte harddisk and 4Mbyte of memory. The
rootimage was initially created from the 0.95a rootimage downloaded from
tsx-11. I used gcc 2.1 (downloaded from tsx) to compile the kernel and
everything works very well. I would like to offer my personal thanks to
Linus and the rest of the development on making such a wonderful tool
available to software community.
The only major piece that still needs to be fit into place is a communications
system to be able to allow the LINUX machine to talk to the rest of our
UNIX(c) machines. I downloaded the Taylor UUCP binaries from tsx-11 and
untarred them onto the machine. I created a /usr/spool/uucp subdirectory
and changed ownership and group to uucp.
For development purposes I have this machine attached to an 80386dx
(Gateway 2000) with 8Mbyte of memory and a 300Mbyte hardrive running XENIX(c)
2.3.3. Serial port 1 (COM1) on the LINUX machine was connected to port 2
of a 12 port Equinox Megaport card. A standard 3-wire connection was
made between the two machines.
I made entries in the Systems and Devices files on the XENIX(c) machine to
allow a direct connection from the XENIX(c) machine to the LINUX machine.
using cu I was able to establish a login from the XENIX(c) machine to the
LINUX machine at 9600, 2400 and 1200 baud. All three speeds provided a
satisfactory connection which I was able to run emacs etc through without
Knowing that there was a valid serial connection between the two machines I
tried establishing a uucico connection from the XENIX(c) machine to the
LINUX machine. It was at this point that problems developed. Using uutry
on XENIX(c) I can watch uucico establish a connection. I made an entry
in the password file on the LINUX machine for a login name of nuucp which
was to start /usr/lib/uucp/uucico. Using ps on the LINUX machine I can
see that uucico starts running.
UUCICO runs through the handshake procedure and announces that the g
protocol is starting. An H message gets sent and from that point onward
I see nothing but alarm messages. Finally after about 12 alarms UUCICO
aborts and announces failure.
From reading this newsgroup I am aware of serial port driver problems due
to the disabling of interrupts during hard-disk activity. The problem does
not appear to be speed related since identical behavior is seen at 9600,
2400 and 1200 bps.
I ran uuchk in /usr/lib/uucp of the LINUX(c) machine and do not see any
real complaints. My Devices and Systems file on the LINUX(c) machine are
very simplistic. According to my understanding the Devices file is not
important in a dial-in environment and the Systems file is simply
arranged so that LINUX(c) knows of the existence of the calling machine.
From what I can tell the problem does not appear to be due to a configuration
problem in either the Devices or Systems file.
I recompiled UUCICO under LINUX thinking that there may be a problem with
the binary. The result of re-compilation was a file of identical length
to the one downloaded from tsx-11. Substituting the new binary resulted
in absolutely identical behavior to the downloaded binary. This seems to
rule out a problem with the binary itself.
The LINUX(c) machine is very lightly loaded. At call-in time the only
things running are update, bash on tty1, gettys on tty2, tty3, tty4 and
ttys1. Swapping is not enabled.
I would interested in whatever comments and suggestions that the net would
be willing to offer. Please use the address in my .sig as this account
is only used for news reading. Thanks in advance for any information which
may be forthcoming.
Dr. G.W. Wettstein
Oncology Research Division Computing Facility
Fargo Clinic / MeritCare
`The truest mark of a man's wisdom is his ability to listen to other
men expound their wisdom.'