Those who know Linux pretty good already won't think much about installing it on a "one-slot-ISA-no-CD" box.
I however -being a newbie to Linux- needed some cups of coffee and half a pack of cigarettes, before I was able to make a decision on the "howto".
Plan A) involved a cable from an IDE Travan Streamer.
It has 2 x 40-pin connectors and 1 x 40 pin male plug. This plug could be used on the original cable to the HD, while the other standard IDE-connector could be plugged into a CD-ROM ... unfortunately this cable is rather short. And I dislike fiddling around with open machines. And the power supply on the PS/2e has no standard DC-plug for another device. Well - that could be bypassed with a second external power supply.
After some time of arranging, plugging and unplugging components I found it -all in all- to unelegant and awkward.
So I changed to the alternative
Plan B) - which is always the plan, when A) turns out not to work or being otherwise unpracticable.
Here I planned to use the IBM TR-adapter and the network to establish a NFS connection to my already running 9595-AMT machine.
I almost got grey hair on that.
I used the original SuSE 6.1 starter disk, which came with the package of 5 CDs and which I thought should do the trick, since the PS/2e is a "generic ISA box" - more or less.
Right so far - but on the attempt to load the Token Ring driver module the FDD choked every time with a read error. I replaced the FDD. No success.
The secret: when the message boot: appears on the PS/2e screen you have to type in floppy=thinkpad - which tells the boot disk to switch to an alternative FDD-controller mode used by the IBM Thinkpads only.
To be honest: I discovered that after I compiled my own kernel, which already included the Token Ring drivers ...
After that I had some "minor" problems with the NFS, because I haven't had the CD-ROM on the host-machine in the exports list:
Assumed you'd installed and started the NFS server ("starting NFS server" during loading) on another machine running Linux already, check your /etc/exports file if there is a line
Of course you need to have the CD-1 from your Linux distribution mounted to the directory /cdrom with using mount -t iso9660 /dev/scd0 /cdrom. The /dev/scd0 might be /dev/cd0 if you are using a Non-SCSI machine as NFS server.
Don't forget to umount /cdrom when you need to swap the CDs ... and to mount the next one, if required during the install.
But that were only small problems then. Afterwards I could install Linux on the PS/2e with no further problems. I decided to leave the Disk Manager on the drive, because the DRPA has over 2.000 cylinders and I wanted to avoid unnecessary fragmented partitioning to keep Linux and lilo within the first 1.024 cylinders.
No problem: the two work fine together.
XF86Config on PS/2e
This XF86Config has been written for the combination 9533 PS/2e with the 9507 10.4" TFT LCD monitor. The settings for the monitor however are generic enough to result in a picture at least on any other screen. They are VGA-monitor settings mainly.
AutoRepeat 500 5
# this line is for a German keyboard !
# see alternative keyboard section below
#Alternative Section for UK-keyboard on US-map:
# Protocol "Standard"
# AutoRepeat 500 5
# Xkbkeycodes "xfree86"
# XkbTypes "default"
# XkbCompat "default"
# XkbSymbols "en_US(pc101)+gb"
# XkbGeometry "pc"
# 640x480 @ 60.1 Hz, 31.5 kHz hsync
Modeline "640x480" 25.25 640 680 776 800 480 491 493 525 +hsync +vsync
# it is important for the X_AGX server to name Chipset and Ramdac !
# theoretical 90 ... but gets flaky after 77MHz usually
# as in the setup
# the COPbase are the last 256 (FFh) bytes of the range
# set under "Video Coprocessor Register"
# here's the clue: must be below 16MB
# standard VGA value
ViewPort 0 0
Some hints for debugging XF86
This chapter basically refers to all machines you wish to run with XF86.
The debugging of the XF86 often takes very long, because the screen stays dark, blanks, returns with garbled characters - or the machine totally crashes.
One method is to pipe the output into a file, using startx -- -bpp 8 >textfile.txt for example. This command starts XF86 in 8bpp mode and redirects the output into the file textfile.txt.
Main pitfall: when the XF86 crashes your machine and you have to "Master Reset" (switch off) the file isn't written and the info is lost. The effort is also wasted, when the screen comes dark, stays dark and you have no clue how to get back to the command line without switching the machine off - or the characters are totally garbled and you have to init 6 or reboot or blindly type shutdown -r now.
It might take very long until you get into the nature of the problem.
If you have a network and the machine connected to it you might use telnet to log on to that machine.
You should have the line ROOT_LOGIN_REMOTE="yes" included in /etc/rc.config - which allowes you to log in to the machine as root from a remote station.
Alternatively you need to have a user defined on your machine other than root and use the "Super User": login with the user via telnet, then type su and use the root-password ... voilá - you are root now ... or at least very close to.
However: it is not necessary for starting the X-Server. You only need to be able to log in from a remote station into the machine - that will do. No need to be root or Super User here. Except you want to be able to edit the XF86Config from another station too.
If you enter startx -- -bpp 8 here and your XF86Config contains e.g. a typo - you will see where it chokes - and why.
If you -for example- forgot to name the chipset for the X_AGX server you get this message:
(**) FontPath set to "/usr/X11R6/lib/X11/fonts/local/,/usr/X11R6/lib/X11/fonts/m
(**) AGX:Valid AGX/XGA Chip type must be specified.
*** A configured device found, but display modes could not be resolved.***
Fatal server error:
no screens found
This is a rather trivial error and you will be able to read it after XF86 returns from the graphic mode - but if the error would have crashed the machine you have had no clue at all.
Attention: the system is no racer. It takes a while until you see the graphic display after the screen blanked. You should immediately stop the X-server when the monitor starts making funny whistling noises or goes totally out of sync. Running a monitor out of sync for longer may result in permanent damage of the monitor (and the video chip in the worst case too !).
To shut down the X-Server locally you need to press the keys [CTRL] + [ALT] + [BACKSPACE] simultanously. Normally that will stop the X-Server from running and return the machine to the command line. But sometimes the screen output is garbled (on all terminals [ALT][F1] to [ALT][F8] ...), the screen stays blank nontheless or the system hangs anyway.
In this case you will have to "Master Reset" the machine or type init 6 blindly.
To shut down the X-Server from telnet you simply press [CTRL]+[C] ... and you are back on the command-line - after some time to shut down the graphics maybe.
Hope this is of some use ... probably not only for running the PS/2e with XF86.