tkined missing tools menu

Mike Thibodeau (miket@hmmh.com)
Tue, 25 Mar 1997 08:45:34 -0500

It turns out that bind was the problem.
I had grabbed the newest release of bind but the symbols had
changed to be posix compliant which caused tkined to barf.

Quick solution:
use bind < 4.9.5
bind 4.9.4p1 works fine following the porting notes

Here is some noise I sent back and forth tracking the problem.
I should have sent this ti the list but didn't *shame*

I darn myself to heck.

Here is the clue that it is a bind library problem:
% tkined1.4.5
ld.so: call to undefined procedure _res_init from 0xef759b5c

The undefined is caused by the lack/renaming of res_init in the new
bind.

I should have spent the time to actually fix the ttnm libraries but
was pulled off to handle some other "important" task...

I had much help from Doug Hughes on this....
What follows is one of the last Emails that we passed on this subject
where I hit the nail as to exactly what was causing the failure.
Someone should incorporate this into the tnm libraries...

volunteers?

From: Michael J. Thibodeau
Date: Thu, Jan 30, 1997 4:45 PM
Subject: Re: make logs
To: Doug Hughes
Cc: miket
> There is a possibility that it has something to do with symbol renaming in
> 4.9.5 though. Perhaps you could try 4.9.4 and see if it works?
>
I am pulling down 4.9.4p1 and will build under that to test it.

> I don't see anything particularly evil in your build sequence. gcc should
> work fine with the bind files (vs sun bundled compiler) if everything
> else is down with gcc too.
>
The bind res libs weren't built with gcc. I built them with gcc but
they still
come out the same.

> The main thing I would do is check res_init.c in the bind/res directory.
> That's where your missing symbol is defined. Perhaps in 4.9.5 they made
> it into _res_init instead. (It was res_init in 4.9.4 and 4.9.3).
> Or maybe gcc is adding an extra _ (to make it __res_init). I've seen
> that before too.
>
cc and gcc both make ___res_init

> do a nm on tnmDns.o and look for res_init
[7:09pm]lexington(235)% nm tnmDns.o | grep -i res_init
U _res_init

> do a nm on res_init.o and look for res_init.
[7:09pm]lexington(234)% nm bind.libs/res_init.o | grep -i res_init
00000000 T ___res_init

> The two should match. Adjust your res_init source code and recompile and
> relink tnm.so accordingly and everything should work okay.
>
Since 4.9.4 works (for you) I am going to go with 4.9.4's res_init
once I get them both side by side I will see what gives.

looking in the CHANGES file for bind-4.9.5-REL I see mention of __
692. [port] fixed up the __ defines (for POSIX.3) in
include/resolv.h.

looking at the source, it looks like the tnm extensions are what should
get
changed so that they comply.

More later....
Thankyou for the starting off point.

-miket

--

-- 

Mike Thibodeau 40 Brook St. Scituate MA 02066 (617)545-2948 mike@cs.umb.edu

And Now...

Mike Thibodeau Network Architect Harris Miller Miller and Hanson Inc. 15 New England Executive Park Burlington MA 01803 Tel.(617)229-0707 Fax.(617)229-7939 miket@hmmh.com

--
!! This message is brought to you via the `tkined & scotty' mailing list.
!! Please do not reply to this message to unsubscribe. To subscribe or
!! unsubscribe, send a mail message to <tkined-request@ibr.cs.tu-bs.de>.
!! See http://wwwsnmp.cs.utwente.nl/~schoenw/scotty/ for more information.