Dear All,
My question was:
Has anyone any idea to the cause of the error reported by DEC ladebug
Debugger? The source code does not make any reference to parsesymscoff.C
and there is no such directory src/libsrc/ladebug on the system.
System: AS 3000/600, DU 3.2b DECladebug V3.0-9.
Error Message:
host> decladebug xvmc
Welcome to the DECladebug Debugger Version 3.0-9
------------------
object file name: xvmc
Reading symbolic information ...readPartialSymbols ignored: inifiles_(File)
done
Assertion failed: realType->enumType() == e_common src/libsrc/ladebug/parsesymsc
off.C 3088
Could not start process for xvmc
No image loaded ... Recovering ...
Segmentation fault
host>
All the answers I receive suggested upgrading to v4.0 which I did and the
problem was solved.
Many thanks to all who replied.
Tim Lucia
Jim Broderick
Olle Eriksson
Lucien Hercaud
Thanks,
Martin.
From: olle_at_cb.uu.se (Olle Eriksson)
Bug in the debugger itself. The file referenced is part of ladebug.
Your version is a bit old. Try Ladebug 4.0-7 instead.
From: Lucien HERCAUD_at_paribas.com
This is the MODULE INSIDE DECLADEBUG where decladebug brakes. Something in
your code broke an assertion made inside the code of decladebug. You can
either try to modify your code and retry to debug or submit the problem to
DEC.
____________________________ Sparateur Rponse ________________________________
Objet : DECladebug (Assertion failed)?
Auteur : alpha-osf-managers-relay_at_sws1.ctd.ornl.gov_at_INTERNET CCROUTER
Date : 21/05/1996 09:38
From: jebroderick_at_CCGATE.HAC.COM
Martin,
I just read your post from a few days ago.
We had the same problems with ladebug when debugging FORTRAN. The
steps you should take to resolve them are:
1) Get the most recent versions of FORTRAN and ladebug from digital.
(FORTRAN 4.0 beta 2 and ladebug v. 4.0-17) These seem to work well
together except for some startup warnings with code containing
BLOCKDATA modules.
2) If your support contract does not cover option 1, an interrim fix
is to rewrite your code to remove any alternate entry points. (i.e.
change ENTRY's to SUBROUTINE's). Old versions of ladebug just
couldn't handle those at all. With out the new FORTRAN you'll still
have some problems, though, especially with core files.
Note that dbx has a complementary set of bugs for FORTRAN, so the
using two together is frequently enough to get by. gdb works well if
you can remember to put an underline character after all your
variables and you don't want to examine common blocks. We never had
trouble with ladebug and C, so f2c might also be helpful.
Good luck,
Jim Broderick
Hughes Missile Systems Company
Tucson, AZ
From: Tim Lucia <lucia_at_zko.dec.com>
Martin,
Sorry for the delay, but I've been on vacation. What you are experiencing with
ladebug (nee DECladebug) is an internal error. src/libsrc/ladebug/parsesymscoff
is the coff symbol table parser inside of the debugger. This problem was fixed a
long time ago. You need to get two things:
DEC Fortran V3.8 or later
DECladebug 3.0-16 or later (note: changed to "ladebug" in 4.0)
Make sure that ALL traces of fortran code are recompiled. This problem is due to
the fact that block data subprograms look exactly like alternate entry points
when compiled with earlier fortran compilers.
This problem has been discussed in comp.unix.osf.osf1 and on alpha-osf-managers
before and is well-known. If you have any further problems, please send mail to
me directly.
Regards,
Tim Lucia
ladebug
(I did most of the fortran support)
--
Timothy J. Lucia Email: mailto:lucia_at_zko.dec.com
Digital Equipment Corporation Phone: 603.881.2615
110 Spitbrook Road ZK2-3Q08 Fax: 603.881.0120
Nashua, NH USA 03062-2698 I'd rather be cycling.
Received on Fri May 31 1996 - 16:29:58 NZST