While ago, just for fun, SUCCESSFULLY compiled JNOS-2.0k on Debian-8.7/64 bit with CFLAGS -m64
This binary is running right now on desktop PC instance of JNOS: 2LPJNS:SP2L - simply flawlessly... Hi!
Ok, that is good! I once tried this with NETCHL and it failed, and I did not want to spend the effort of going through all code to make it 64-bit secure, and this is a waste for such a program anyway (it does not require large integers or pointers that can access more than 2GB). So for NETCHL the -m32 flag is mandatory.
Of course, with Debian8.7 there still are several compiler options (4.6 - 4.9 and maybe more) so there may be some issue with a different version of the compiler, I have seen that happen in the past.
Otherwise I have no clue what could be happening, we would need at least the stack backtrace and maybe more.
Rob
Rob / PE1CHL
Debian-8.7 distro version of GCC is: gcc version 4.9.2 (Debian 4.9.2-10)
Best regards.
Hello All,
More to the saga of my JNOS2.0K netrom problems here.
The thought was that there is a problem compiling JNOS2K on Fedora so for the last few days I have been loading the latest Debian on another PC. Along the way I have loaded Debian 4 times, groan! The first three were using the GNOME desktop and yes, no display available after load and reboot was off the live disc. (Old video card) Using KDE on that PC now ok.
But after compiling JNOS2K on Debian, it again failed. It seemed to compile just fine though. So then installing it all again using the "installer", I thought I had it fixed only to find version H came with the installer. Of course it worked fine. But after loading my JNOS2K into the structure it again crashed.
Noticed on both machine that if I add a manual "netrom route" and do a netrom bc xx, it crashes instantly'. Otherwise if I leave it to do its own nodes broadcast, it crashes then. (Both Fedora and Debian builds)
I have tried compiling with the original config.h from the GZip file and changing some bits but all builds still crash.
If you are wondering, I am not loading encap routes, axip links or local routing, it still crashes. I have also used various directory configurations. I have used the default built-in netrom parameters also. Currently using my "home" director path on Debian and Root Path on Fedora.
Versions H & I and then a K compiled in last December work fine but not K today.
So it would be great if someone with more knowledge than I would download the current JNOS. 2K GZip file from Langelaar.net (don't use your original one) and see if it works for you. Thanks.
Best of luck Rob vk1kw
-----Original Message----- From: 44Net [mailto:44net-bounces+vk1kw=netspace.net.au@hamradio.ucsd.edu] On Behalf Of SP2L Sent: Thursday, March 16, 2017 8:03 AM To: 44net@hamradio.ucsd.edu Subject: Re: [44net] FW: JNOS2.0K and Netrom problem
(Please trim inclusions from previous messages) _______________________________________________
_________________________________________ 44Net mailing list 44Net@hamradio.ucsd.edu http://hamradio.ucsd.edu/mailman/listinfo/44net
Regrettably, I don't have a way to test JNOS here, so this is just a guess:
A faint possibility is that there is something in the latest version of the source code that is tripping a bug in the compiler. This is not very likely, but one thing I have found helpful in debugging programs that are acting mysteriously is to change optimization when compiling. If the Makefile sets the CFLAGS to -O, try -O0 instead.
This can also help if you are running the program under GDB in that it can make the fault a little easier to track down.
Best wishes - Brian
On Sat, Mar 18, 2017 at 08:06:57PM +1100, vk1kw wrote:
More to the saga of my JNOS2.0K netrom problems here.
To whom it may concern.
Puzzled by reported issue with JNOS Netrom crashes did some investigations and got POSITIVE result, Hi! Two files in source tree of JNOS contain remnants, sequence of CRLF, after being edited under _NON_ Linux systems.
Files affected: chkstk.c wlauth.c
It is necessary to treat both files by, for instance on Linux, program like dos2unix.
Then compilation and further usage of jnos binary is PERFECTLY seamless - NO crashes on NET/ROM broadcast!
This bug pertains also latest version - jnos-2.0k.1
Verified on Debian-8-7/32bit and Debian-8.7/64bit
Above reported also directly to Maiko VE4KLM.
Best regards.
Greetings.
Unfortunately, need to retract previous assumption. My verification was not profound enough. Must be something else causing issue in question.
Best regards.
Exactly along the lines of Brian's post, I recently had a report of LInpac crashing at started up on Glibc 2.24 / GCC 6.3.0 based systems (Debian 9.0/Stretch, Debian 10.0/Sid, and Arch Linux). This was really strange as this program has been working just fine call older generations of Linux distros, etc. The initial crash was always consistent but we found over four different workarounds to get it to successfully run which did not make any sense. Ultimately, it turned out that the new GCC was behaving differently on memory release operations compared to say GCC 4.9.2 (found in Debian 8.0 / Jessie). A few code changes were require to stop using the library call "delete[]" and now use "free ()" solved the issue. Definitely a pain but this is one of those "progress" pains that all software developers have to pay.
As to how to better troubleshoot this Jnos issue, you need to run a gdb backtrace. I just checked Jnos 2.0K and it leaves in debugging objects by default (unnecessary actually). Once compiled, run Jnos via gdb, "run" the processand once it crashes, request a backtrace. This is a decent, non-complicated example:
http://wiki.freepascal.org/Creating_a_Backtrace_with_GDB
From there, take this backtrace detail to the Jnos support email list and Maiko.
--David KI6ZHD
-
Brian Kantor -
David Ranch -
Rob Janssen -
SP2L -
vk1kw