Return to BSD News archive
Newsgroups: comp.unix.bsd
Path: sserve!manuel!munnari.oz.au!uunet!usc!sol.ctr.columbia.edu!e40-008-9.MIT.EDU!vikki
From: vikki@e40-008-9.MIT.EDU (Vikki King)
Subject: 3com AUI/kernel bld. WORK!
Organization: Columbia University Center for Telecommunications Research
Message-ID: <1992Sep10.093709.10686@ctr.columbia.edu>
Sender: news@ctr.columbia.edu (The Daily Lose)
Date: Thu, 10 Sep 1992 09:37:09 GMT
X-Posted-From: e40-008-9.mit.edu
X-Posted-Through: sol.ctr.columbia.edu
Lines: 93
Hi Everyone!
Well, I have finally succeeded at getting my kernel to build when I
want it to AND I have gotten my 3c503 card to talk out its AUI port - Yayyy!
I had to overcome the problems with the Makefile and vers.c and also how to
change if_ec.c to make the AUI port come alive. BTW, thanks to all who
helped me get this far! Ok, here's what I did:
1. First, to fix the sources to use the 3c503's AUI port, go into the
file /usr/src/sys.386bsd/i386/isa/if_ec.c. At the end of line number
277 and change the code from:
E33G_CNTRL, 2);
to
E33G_CNTRL, 0);
You may want to save an unmodified version of if_ec.c for purist sake
or at least make note of the change in the file in case you ever have
to undo it.
2. Next, to fix the problem of the kernel not building because of
the undefined symbol thing, and to get my if_ec,c changes in place, I got
a copy of cgd's fixed Makefile from agate.berkeley.edu and used the patch
program to update my copy of Makefile.i386.
3. Then, I copied the GENERICISA file to one called BLUEFISH (or whatever
floats your boat). I then edited BLUEFISH to contain just the drivers
that my system needs - I got rid off all the other ethernet drivers and
the adaptec/QIC-02 drivers - I don't got none of those. After that, I
issued a
# config BLUEFISH
command.
4. Now, a little finagling was necessary. I went to the compile/BLUEFISH
directory and had to copy the file ../GENERICISA/vers.c and the file
../GENERICISA/version into compile/BLUEFISH. I then issued a:
# make depend
and a:
# make
command.
5. When it was all done, there was still a load problem because vers.o
couldn't be found. I peeked in Makefile again, found that using the
command:
# make newvers
would create vers.o and went ahead and did it. When it was done, I did
another # make depend/# make and like magic, I got a new kernel!
Finally, the above is a summary of what I have found out so far. I actually
had a bunch of trials and failures that I left out of the summary. One more
thing that I found was important is that whenever you make a change in any of
the source anywhere, run a # make depend before running # make. I actually
had the AUI thing figured out quite a few tries before I actually knew it
because I was just doing a plain old # make after I'd tweak something.
Now, for something related:
After ifconfig'ing ec0 and successfully ftp'ing/ping'ing a host on our net,
I am having problems trying to access an NFS server. The NFS server is a
netware 3.11 box with Netware NFS v1.1. I can get the server volume to mount
on the 386bsd filesystem and running df gives an accurate report of the space
available, but I am unable to do anything in the directory where the NFS
volume is mounted. Trying to do an ls or cp'ing something into that directory
just gives me a permission denied error. Doing an ls -l of the mount
directory after I've mounted the volume shows the privledges as being:
drwx------ 4 root 512 Sep 9 16:30 mcet1/
It would seem as if I had the proper privledges, right? Last, if I explicitly
try to mount the NFS volume by hand with the command line:
# mount /bsd386/@mcet1: /mcet1 (I know, I know, it should be /386bsd/ :-) )
I get a message that reads "Can't get net id for host". I can't figure out if
I've got something set up incorrectly on the Netware NFS hosts's side or if
the problem is somewhere in how I've configured the 386bsd system. Anyone
want to offer any suggestions?
-John
jackson@a1.mec.mass.edu