GeezerVison is showing linux-visws.sourceforge.net — Linux for SGI Visual Workstations
Linux for SGI Visual Workstations
Linux 2.2.10 + Red Hat 6.0
NOTE: These instructions need to be updated.
Updated: July 28, 1999
Step 3: Post-installation cleanup
The Red Hat install program does not know anything about Visual Workstations.
This document does.
Follow it closely
Booting the Linux RX kernel
- find Floppy #1
If you followed the previous
instructions precisely and have nicely not futzed with the machine at
all the screen now has a nasty looking box saying:
"Failed to start system: Unable to load bootfile".
Click "Continue"
If you do NOT see this message just press the reset button
and hit ESC to get you to the "WELCOME" screen of 6 buttons.
Insert Floppy #1 now.
- Start System
Recall that we had left "Linux RX" the Default
Within a couple dozen seconds the penguin reappears in
the upper left corner with some grotty kernel boot gunk.
- find Floppy #3
Floppy #3 is the "rescue.img" floppy.
Please insert it in the floppy drive now.
Press the Enter key and you'll instantly see
"RAMDISK: Compressed image found at block 0".
Wait for the floppy drive light to go out (30-ish seconds).
A classic single user prompt is now on your screen.
Type "mount" and/or "ls" and revel in the Unixification of
your Visual Workstation!
You are running now in a ramdisk root.
For some reason neither Backspace nor Delete do the right thing.
Neither does ^H --
^U kills a whole line.
But, don't worry, we're moving on to a slightly friendlier environment.
- mount real root
- mkdir /r
- mount /dev/your-ext2fs-root-partition /r
Did you remember what disk and partition you appointed '/'?
- chroot /r
You'll see "bash# ".
You're now running on the real root file system -- just FYI.
And NOW the ^H does work as the delete key.
WARNING: You are now running as super-user and have all
powers necessary to destroy your system.
- Floppy #4
Finally... last floppy.
- swap in Floppy #4
- mount -t msdos /dev/fd0 /mnt/floppy
- ls /mnt/floppy
You should see the tar image we asked you to
put on this disk back in the
pre-install instructions.
Extract everything from that tar image:
- cd /tmp
- tar zxvf /mnt/floppy/f4jul28.tgz
(GNU tar has gzip built-in -- the z option --
for those of you used to IRIX).
Verify that it's correct:
- cd f4
- sum *
Compare to this.
- install bootloader
To not conflict with the Red Hat /boot directory
I temporarily mount the boot partition here as "/b".
(You don't actually have to have this permanently mounted).
- make boot partition FAT
If you are using an existing FAT partition --
especially one that is used to boot any other
OS on your machine --
SKIP these four steps and go on to the "mkdir".
-
/sbin/mount /dev/cdrom /mnt/cdrom
-
rpm -i /mnt/cdrom/RedHat/RPMS/mkdosfs*.rpm
-
/sbin/umount /dev/cdrom
-
/sbin/mkfs -t msdos /dev/fat-boot-partition
- mkdir /b
- /sbin/mount -t msdos /dev/fat-boot-partition /b
Here is where you remember the boot partition.
- cp arclx.exe /b
This assumes slavish adherence to the instructions
in that you're still cd'ed to /tmp/f4 and have
the contents of the tarball in the current directory.
- /sbin/umount /b
- install Visual Workstation generic kernel
cp la2210.vw /
This is really just temporary until you grab the
Visual Workstation Linux kernel patch and build
a proper kernel in a later step.
- frame buffer gunk
cp fb.modes /etc/fb.modes
- X server gunk
- XF86Config gunk
If you have a regular CRT monitor pick the ".crt" file.
If you have a 1600SW flat panel pick the ".fp" file.
- cp XF86Config.crt /etc/X11/XF86Config.vw
- cp XF86Config.fp /etc/X11/XF86Config.vw
mknod /dev/hidbp-mse-0 c 10 32
Human Interface Device, Boot Protocol...
it's a USB thing.
This is how the mouse appears to
the X server.
You'll see the same dev name in this XF86Config.
It MUST MATCH EXACTLY for X to start.
Those are 2 single dashes ('-') and a zero ('0').
- some cleanup...
For optimum neatness and best results:
- cd /
- umount /dev/fd0
- exit
You're now back to just the '#' prompt.
- umount /r
- press the reset button
And hit ESC when the PROM comes up
- now: reboot!
- create your Linux Boot Selection
- bonk on Startup Settings
- bonk on New
- Load Identifier
Make something up.
"Red Hat 6.0 + Linux 2.2.10" or some such.
- OSLoader
arclx.exe
- OSLoadFilename
Here is where you remember
1) your Linux system/root partition,
and 2) the mapping between hdXY type names
and multi(i)disk(j) type names
and 3) the name of the generic kernel.
For example,
if your root partition is hda2 (an IDE disk)
and the kernel is called "/la2210.vw" type this:
multi(0)disk(0)rdisk(0)partition(2)/la2210.vw
Or, if your root partition is sda2 (a SCSI disk)
and the kernel is called "/la2210.vw" type this:
multi(2)disk(0)rdisk(0)partition(2)/la2210.vw
(Yes, that "multi(2)" stuff is wierd).
- OSLoadPartition
Flip through the choices until you see the
one you just copied arclx.exe into.
(/dev/hda1 is known as "IDE 0 Disk 0 Partition 1" and
/dev/sda1 is known as "SCSI 2 Disk 0 Partition 1")
The Partition number is _physical_ numbering, not
the logical/extended names.
The first partition on the disk is '1', the next
is '2', etc.
- System Partition
Don't worry about it
- OSLoadOptions
This must match the disk/partition from
OSLoadFilename (making the wild leaping assumption
that you put your kernel in the root file system)
FOR EXAMPLE,
if your root partition is hda2 then type:
root=/dev/hda2
NOTE: YOUR ROOT PARTITION MAY BE DIFFERENT!
Also, the kernel does NOT automatically probe for the
amount of RAM installed in your system.
It defaults to 128MB.
If you have more than 128MB you can use the "mem=xxxM"
option here on the OSLoadOptions line to make use of
all of your memory... mem=512M if you have 512MB, etc.
The Hardware Inventory "button" on the "6 button"
boot firmware menu will report the amount of RAM in your system.
- Default
Make this your default.
Save&Exit
- BOOT LINUX!
Bonk on Start System
You'll see a nice blue box (not to be confused with "blue screen" :-)
yakking about ARC boot this and that and some partitions and kernel
file name that should be numbingly familiar to you by now.
Then, you'll see that little penguin and the grotty boot stuff...
Eventually, you'll see "Red Hat Linux release 6.0 (Hedwig)"
and a login prompt.
Well... login!
... as root ... remembering the root password.
NOTE: See the known bugs list before you
run around willy nilly hacking /etc/passwd directly or using
passwd to change passwds.
- X setup
To do anything moderately visual you install and setup some X stuff.
If your Red Hat 6.0 CD isn't already in the CD drive get it in
there now.
- mount the CD
mount /dev/cdrom /mnt/cdrom
This presumes /dev/cdrom is a symlink to the right place
(most likely /dev/hdc on a factory-fresh system).
- remove bad X stuff
Remember when we selected the "Mono" X server just
to get through the install procedure?
We remove that now.
rpm -e XFree86-Mono
rm /etc/X11/X
- install good X stuff
rpm -i /mnt/cdrom/RedHat/RPMS/XFree86-FBDev-3.3.3.1-49.i386.rpm
ln -s ../../usr/X11R6/bin/XF86_FBDev /etc/X11/X
- configure X
cp /etc/X11/XF86Config.vw /etc/X11/XF86Config
(We copied the XF86Config file from the floppy
to this indirect holding name just to make sure
nothing "smart" trashes it for us)
- START X!
startx
The screen will clear and repaint and you'll get
that "wonderful" "X".
If you had selected GNOME way back during Red Hat install
you'll eventually see that footprint thing.
Point and click with the USB mouse.
Aim maybe for the tty icon and make sure the USB kbd
types the right stuff.
If you DO NOT see the X and instead see some teensy
green text on black background this means that X server
startup failed.
You're simply staring at the virtual terminal running
on a differently dimensioned framebuffer.
Type "clear" very carefully and you'll see less
green crawly stuff, etc.
If you configure the network and run startx over from
an rlogin session you'll more clearly see the complaint
from startx.
Here are the most common reasons startx fails:
- CONGRATS!
You've succeeded!
One more step if you really are so inclined:
building a kernel from source.