Re: [CAD_CAM_EDM_DRO] Help Linux Woes Help!!!

> So, I have RedHat 5.2 installed off the CD together with XFree86 and it
> seems to work and plays games OK - first stage in pacifying the wife!

Did you use Workstation, Server, or Custom? I usually use Custom and when it
presents you with the list of packages to install, I scroll down to the end
of the list and pick 'Everything'. If your disk is 1GB+ it will all fit. You
can always erase the Japanese version of the documentation later...

> I downloaded the kernel_source and kernel_headers files since I couldn't
> find them on the CD and I have installed them using the package manager
> as in Tim's message of 9th Jul. I also downloaded the RTLinux
> Release9J.tgz file and copied it into /usr/src and (un)tarred it. Now
> comes the difficulty. When I try to run 'patch -pl <
> /usr/src/kernel_patch' I get the error message 'bash: patch: command not
> found'

Are you logged in as 'root'?

> I have the feeling I should rebuild the kernel or something but I am
> confused by the variety of documentation I now have, most of which seems
> to tell me to run 'make config' or some such, however, when I try this I
> again get the error 'bash: make: command not found'

Either 'make' is not installed (hard to believe) or you're not logged in as
root.

> One of the docs I read says something about deleting everything from the
> /linux directory but, not knowing enough about the way the system works
> yet, I baulked at this.

Don't do this yet. The only reason to get rid of everything under
/usr/src/linux is if you have goofed up some of the source files by applying
the wrong patches and need to start over. You are not in that situation.
Simply put, the files in /usr/src/linux are the source files for the kernel.
You don't need them to run your system, only to make new kernels. The kernel
you are booting is called out in a section of lilo.conf, i.e.:

image=/boot/vmlinuz

Even after you recompile your kernel you need to copy the new kernel file to
the directory (and filename) specified in lilo.conf. We do this by:

cp arch/i386/boot/zImage /boot/vmlinuz

> The computer is switched on grumbling to itself and I don't know whether
> I can switch it off without bu..ering something up or not. Can some kind
> soul please put me out of my misery and explain in idiot language how to
> get myself out of this mess so I can sleep peacefully?

You should be able to switch off and on without trouble at this point since
you haven't edited lilo.conf and run lilo to change the name of the kernel.
I'm going to paste the complete instructions at the end of this message so
that you'll have them to refer to. I've tried to fix the bugs in them that
Tim found, although I haven't added his suggestion to allow booting the old
kernel by having multiple kernels listed in lilo.conf. This is a good
suggestion, I just haven't put it in yet.

Good Luck and don't hesitate to ask for more help if needed,

Matt
-----
_EMC Software Installation Instructions_

These instructions assume you have Redhat Linux 5.2 already installed and X
Windows correctly configured.

1. Update the Linux kernel headers and sources. These updates are needed for
the RT-Linux patch to work properly. The updates are at:

http://www.redhat.com/swr/ftp.redhat.com/redhat/updates/5.2/i386/

as Red Hat Package Manager (RPM) files. Download the kernel-headers and
kernel-source packages as well as the update for the X server you are using
and anything else that strikes your fancy into a directory of your choice. I
have always downloaded and updated everything.

As user 'root', cd to the directory containing the update RPM files and do:

rpm -Uvh kernel-headers-2.0.36-3.i386.rpm
rpm -Uvh kernel-source-2.0.36-3.i386.rpm
rpm -Uvh etc... for all the remaining updates you downloaded

When updating the kernel headers you will probably get an error message
saying something like "script failed to complete". I have always ignored this
error and there was no problem. You may get a message about "dependencies".
This means that the RPM you are trying to update depends upon another update
being installed first. The dependencies will be listed and you must update
those RPMs first. Usually, I just do:

rpm -Uvh *.rpm

repeatedly until I get messages indicating that all the RPMs are updated and
nothing more can be done.

2. Install the RT-Linux patch, available at:

ftp://ftp.isd.cme.nist.gov/pub/emc/emcsoft/release9J.tgz

Download this file and place it into /usr/src.

As user 'root', do:

cd /usr/src
tar xzvf release9J
cd linux
patch -p1 < ../kernel_patch
cd ../rtl
make all

3. Recompile the kernel.

As user 'root' from a terminal window in X, do:

cd /usr/src/linux
make xconfig

You need to be running X Windows for the 'make xconfig' line to work. If you
don't want to run X, you can do 'make config', and configure in text mode.
You would do well to take some time with this process and eliminate any
modules or kernel support for hardware you don't have or features you won't
use or don't need. This will SPEED THE BOOT PROCESS and DRASTICALLY REDUCE
THE TIME NEEDED TO COMPILE THE KERNEL! Failure to heed this advice probably
won't cause any problems, but the command 'make modules' can take more than
an hour to execute if many modules are to be built. Configure your kernel to
your liking, or leave it as-is by clicking 'Save and Exit'. Now continue with
the kernel compile:

make dep
make clean
make zlmage
make modules
make modules_install
cp arch/i386/boot/zImage /boot/vmlinuz

If there were errors that prevent the kernel from being compiled, the file
'arch/i386/boot/zImage' won't exist and thus can't be copied. If you have
trouble go back to the beginning of this step, reconfigure, and try again.

4. Edit /etc/lilo.conf so that it looks somewhat like this:

boot=/dev/XXX <-- leave xxx as is, e.g., hda1
map=/boot/map
install=/boot/boot.b
prompt timeout=50
image=/boot/vmlinuz <-- change this to /boot/vmlinuz
label=rtlinux
root=/dev/XXX <-- leave XXX as is, e.g., hda1
read-only
append='mem=XXm"<-- XX = the number of MB of RAM you have minus 1

5. Run LILO to set up your machine to boot the new kernel.

As user 'root' do:

/sbin/lilo

and then reboot.

6. Now you need to get the EMC software from:

ftp://ftp.isd.cme.nist.gov/pub/emc/emcsoft/linux_2_0_36/

and put it into /usr/local/nist. This is available as one big file named
emc-DD-MMM-YYYY.tgz, where DD = the day, MMM = the month, and YYYY = the year
that the file was created, or 4 smaller files that will each fit on a floppy.
To use the small files you can copy them onto the disk from floppies and then
use 'cat' to combine them back into one file.

7. Unpack the files and install the software.

As user 'root', do:

cd /usr/local/nist
tar xzvf emc-DD-MMM-YYYY
./install

This will build all the programs and put all the files where they belong.

8. Configure the script that runs the EMC and the .ini file that controls its
operation.

This is a big subject in itself, however you can try out the software by
getting these two files:

http://www.erols.com/mshaver/run.64step378
http://www.erols.com/mshaver/64stepper.ini

and putting them in /usr/local/nist/emc. These files assume you have followed
the above instructions about where to put the software. They are also
specific for machines with 64MB of RAM. If you have a different amount of RAM
you must edit these files to replace all instances of the string 0x3F00000
with a value such as:

0x0F00000 for 16MB systems
0x1F00000 for 32MB systems
0x2F00000 for 48MB systems
0x3F00000 for 64MB systems
0x4F00000 for 80MB systems
0x5F00000 for 96MB systems
0x6F00000 for 112MB systems
0x7F00000 for 128MB systems
etc..

These files also depend upon there being a parallel port at address 378h.
This corresponds to LPT1 in DOS and if you only have one port it will most
likely be this. If not you must edit these files to replace all instances of
the string 0x378 with 0x278 or 0x3BC, whichever is appropriate. Be sure to
disconnect anything plugged into the port you use since this version of the
EMC toggles bits to run stepper motors and could cause "interesting" things
to happen to printers. The advantage of trying out the EMC using these files
is that there is no other hardware required. To start the EMC:

As user 'root' from a terminal window in X, do:

cd /usr/local/nist/emc
./run.64step378

Of course if you edited the files you may also have changed their names so
use the appropriate command. Note that the 'run.' script refers to the '.ini'
file explicitly by name, so if you change the name of the '.ini' file you
will need to edit the 'run.' script to match.