Trouble re-linking kernel on SCO Openserver 5 with Xe 8 board

Am installing Xe 8 multi-port on SCO Openserver 5.0.5
Installed latest MPI version 4.2.1 and received the following error when attempting to relink the kernel for
SCO Openserver 5.0.5

k386ld fatal: Symbol referencing errors. No output written to unix.
ERROR: can not link-edit unix
idbuild: idmkunix had errors.
^GSystem build failed.

Would greatly appreciate assistance.

You should check with SCO regarding the kernel link errors.

The entire error message is shown below:
undefined symbol first referenced in file

pcxx_nbios /var/opt/K/SCO/link/1.1.1Eb/etc/conf/pack.d/pcss/Driver.o
pcxx_nbio
/var/opt/K/SCO/link/1.1.1Eb/etc/conf/pack.d/pcss/Driver.o
pcxx_ncook
/var/opt/K/SCO/link/1.1.1Eb/etc/conf/pack.d/pcss/Driver.o
pcxx_cook
/var/opt/K/SCO/link/1.1.1Eb/etc/conf/pack.d/pcss/Driver.o

i386ld fatal: Symbol referencing errors. No output written to unix.
ERROR can not link-edit unix
idbuild: idmkunix had errors.
^GSystem build failed.

I’ll check with SCO but was just wondering if I could be making an error with an incorrect memory address or some other parameter. The interesting thing is that an earlier version of MPI worked fine with the DigiCHANNEL PC/x that I am replacing. The AcelePort 8e was recognized by the older version of MPI and the kernel relinked but would not work since it requires an MPI of 4.0 or above.

Try removing MPI and relinking the kernel by hand to see if the linking errors still occur. If so, then it would point to some sort of kernel/link-kit corruption.

Manual MPI removal and kernel relink instructions:

http://www.digi.com/support/kbase/kbaseresultdetl.jsp?id=261

An incorrect I/O address would be a problem, but I don’t think it would generate that type of error. That error had to do specifically with the kernel.

Since you’ve removed the driver and were able to successfully re-link the kernal manually, I’d re-download the driver (make sure you download it as a binary file rather than ascii) then attempt re-installing it after checking/verifying the DIP switches on your Xe card so you know what I/O address the card is set to before installation.

I’ve removed MPI (both automatically and manually) and the kernel re-links without a hitch. I don’t know if I mentioned this before but this error will occur on both of the SCO boxes that I’m creating. One of these is an old SCO install and the other a new SCO install.

I still haven’t conquered this problem. I know I’m showing my ignorance here but when I copy the files from the floppy in order to install MPI via custom do I use the following command? doscp /dev/rfd0135ds18 /digi and then of course copy each file to their respective VOL. file before I install via custom? I can’t seem to transfer the files to a cdrom otherwise I would try that instead.

Download the “single file install” driver from this page:
http://www.digi.com/support/productdetl.jsp?pid=1327&osvid=40&tp=1&s=7

Copy it over to CD, then from CD over to your SCO server. Once on the SCO server, follow the guidance in this article to install the driver (once any previous installs have been removed completely, kernel relinked, and server rebooted):

http://www.digi.com/support/kbase/kbaseresultdetl.jsp?id=914