-----FW: Re: compile kernel and apps, what's up? (fwd)-----

Date: Sat, 08 Aug 1998 05:59:58 -0700 (PDT)
From: Brian Wildasinn <bwildasi@csulb.edu>
To: (Bill Studenmund) <wrstuden@loki.stanford.edu>
Subject: Re: compile kernel and apps, what's up? (fwd)


On 08-Aug-98 Bill Studenmund wrote:
> Took me a bit to notice the extra X in your account name...

Ooops, even more strange is that xfmail can't send a  reply to sender if that
person had from:<bwildasiX@csulb.edu> in his or her header file. And this
header file cannot be shown or changed in the received original or the 'reply
to sender' text window. Sooo, so much for anti-spam. I couldn't even use my
own 'reply to sender sender' message!  

>> > Maybe
>> > that's why a new kernel compile hasn't completed yet. Here's my session
>> > thus
>> > far. The script was only run after gcc bailed out, so there's not a
>> > zillion 
>> > lines.
>> > http://www.csulb.edu/~bwildasi/unix/netbsd/kernel-typescript
>> > http://www.csulb.edu/~bwildasi/unix/netbsd/Q700-conf
>> > http://www.csulb.edu/~bwildasi/unix/netbsd/Q700-compile-Makefile
>> 
>> I'm not sure what the problem you're having is, but the errors you
>> are getting are at the link stage. All your source is from 1.3.2, isn't it?

I wasn't sure myself, so i went back to the netbsd-1.3.2 soruce tree at
ftp.netbsd.org and downloaded again the 
ftp;//ftp.netbsd.org/pub/NetBSD/NetBSD-1.3.2/source/sets/syssrc.tgz, 

>> > Basically I've done 
>> > 
>> > 'make depend && make && make install' in /usr/src/usr.sbin/config/
>> > 'cp GENERIC Q700' in /usr/src/sys/arch/mac68k/conf/
>> > 'config Q700' in /usr/src/sys/arch/mac68k/conf/
>> > 'make includes' in /usr/src/sys/arch/
>> > 'make depend && make' in /usr/src/sys/arch/mac68k/compile/
>> 
>> That should do it, assuming you meant compile/Q700 as Ken noted.
Yep, that's right.
>> 
>> > ps, say is there any takers on why include files aren't being recognised
>> > by
>> > gcc?
>> 
>> What exactly was the error?
See first error, called 'adb.o: ...' at the top of

http://www.csulb.edu/~bwildasi/unix/netbsd/kernel-typescript

and here's what I found regarding the 'undefined symbol' of adb.o:

http://www.csulb.edu/~bwildasi/unix/netbsd/adb.o-typescript

It appears that some header definitions are 'Undefined symbol' to the gcc
compiler when infact the 'undefinded symbol' is right there in the header file,
for example the adb.o error mentioned above, which is the first error mentioned
in the
http://www.csulb.edu/~bwildasi/unix/netbsd/kernel-typescript .

Or, i am probably missing something? 

To me this is similar to local and global include files, ie if IIRC, if the
header is '#include <>', gcc searches global include directories, whereas if the
header is '#include ""',  gcc looks in the same directory of the *.c files to
get the *.h header file. Or maybe I'm confused, which dosen't surprise me!
Anyway, all includes have the <> for global include directories, if that is in
fact the case.

I've been running into similar problems compiling xwindow apps, the Xwindows
distribution for netbsd-1.3.2 has /usr/X11R6/include/X11/*.h files that can't
find their neighbors in the same directory as themselves. My work around was to
hardcode the '#include <>' lines to '#include ""' for the offending, or actually
'missing' X-window header files. When compiling X apps, the headers are "can't
fine '#include <X11/*.h>, whereas when compiling non-windows apps some of the 
*.c variables are 'Undefined symbol', which are clearly header file defined
variables. No follow up on that one from the port-mac68k list yet. Any ideas?

Any help is appreciated!

Brian
bwildasi@csulb.edu


----------------------------------
E-Mail: Brian Wildasinn <bwildasi@csulb.edu>
Date: 08-Aug-98
Time: 05:59:58

This message was sent by XFMail
----------------------------------

--------------End of forwarded message-------------------------

----------------------------------
E-Mail: Brian Wildasinn <bwildasi@csulb.edu>
Date: 08-Aug-98
Time: 07:09:02

This message was sent by XFMail
----------------------------------