Game Development Community

Redhad Segmention fault

by Nick Carnevalino · in Torque Game Engine · 04/21/2002 (2:44 pm) · 11 replies

um .. redhed should of been RedHat ... hehe


# ./torqueDemo_DEBUG.exe -console -dedicated -mission fps/data/missions/test1.mis
Segmentation fault (core dumped)




i tried with and without -console
i got no console .. no nothing .. just Segmention Fault .. i dont care about the GUI i just want a dedacated server. .. oh ya ... compiled from dev-stable grabed via cvs yesterday ( april 20 ) with
# gcc -v
Reading specs from /usr/lib/gcc-lib/i386-redhat-linux/2.96/specs
gcc version 2.96 20000731 (Red Hat Linux 7.1 2.96-98)
#make -v
GNU Make version 3.79.1, by Richard Stallman and Roland McGrath.
Built for i386-redhat-linux-gnu


Linux Distribution Red Hat 7.2
Kernel Version 2.4.7-10
CPU Type & Speed AMD Alathlon 800)
RAM 192MB)
Video Card & Driver Version - NVIDIA GeForce 2 version ( no clue on version but i dont run X much anyway )
i run a nightly up2date cron so i have the latest rpm's off of redhats site ( lastnight i noticed it updated nasm acctully )

Core from the Segmention fault:
This GDB was configured as "i386-redhat-linux".

Core was generated by '/root/projects/torque/example/torqueDemo_DEBUG.exe -console -dedicated -mission'.
Program terminated with signal 11, Segmentation fault.
#0 0x400095fd in ?? ()
(gdb) where
#0 0x400095fd in ?? ()
#1 0x4000b9ec in ?? ()
#2 0x40002af5 in ?? ()
#3 0x40010197 in ?? ()
#4 0x40002441 in ?? ()
#5 0x40002233 in ?? ()
(gdb)
NextNext Thread
PrevPrevious Thread
More from Torque Game Engine


Torque Owner John Quigley
#1
04/21/2002 (2:53 pm)
In your mk/conf.UNIX.mk file, does it say

ASMFLAGS =-f elf -dLINUX

or

ASMFLAGS =-f aout -dLINUX

Torque Owner Nick Carnevalino
#2
04/21/2002 (3:05 pm)
my conf.GCC2.LINUX.mk file ( their is no unix one but i know this is the one used ) has
ASMFLAGS =-f aout -dLINUX
in it ... should i try elf or are you just gathering info?

Torque Owner John Quigley
#3
04/21/2002 (3:08 pm)
That nasm flag is wrong; it should be elf, not aout. Your makefiles are out of date. I suggest you do the following to update.

cd torque
cvs update -r Linux_Beta

Then do a full clean rebuid. Don't update to the HEAD, it doesn't work right now.

Edit: or you can just try elf, but there might be other problem-causing out of date stuff in your tree.

Torque Owner Nick Carnevalino
#4
04/21/2002 (3:32 pm)
re-compiling now ... what i did yesterday was grab the dev-stable from the cvs and copied it to a safe place ...just now i compleatly deleted my torque dir and copied the dev-stable over then did the update ... it seems to be going ok ( i can feel the HD chruning under my feet ( i use the server as a foot rest sometimes )) will see how it goes :-)

Torque Owner Nick Carnevalino
#5
04/21/2002 (3:41 pm)
:-) ! .. looks like it's working now ... somthing about the terrain being wrong but i can deal with that :-)

Torque Owner Nick Carnevalino
#6
04/21/2002 (3:56 pm)
*grumble* it worked once ... not after i stoped and re-started the server

This GDB was configured as "i386-redhat-linux".
Core was generated by './torqueDemo_DEBUG.exe -console -dedicated -mission fps/data/missions/waterWorl'.
Program terminated with signal 11, Segmentation fault.
#0 0x400095fd in ?? ()
(gdb) where
#0 0x400095fd in ?? ()
#1 0x4000b9ec in ?? ()
#2 0x40002af5 in ?? ()
#3 0x40010197 in ?? ()
#4 0x40002441 in ?? ()
#5 0x40002233 in ?? ()

Torque Owner John Quigley
#7
04/21/2002 (4:11 pm)
If it works once but crashes later, that is Extremely Weird. Make sure you're getting the right core dump; this one has the same trace as the last one, which is strange. You could try removing the ~/.garagegames/torqueDemo_DEBUG.exe directory, and then restarting it. This is where the output files (dso and friends) get created.

Torque Owner Nick Carnevalino
#8
04/21/2002 (4:39 pm)
ok it gets weirder ... just for fun i ran make in the torque dir .. it grabed the 5 or so debug.a files from the previous make and linked the torquedemo_DEBUG.exe ... this time it works and i cant NOT make it work .... maby the privous make didnt quite finish everything when it linked? *shrug* .. possalby old files in the tmp?

BTW their a list of basic commands for the colsole? .. i think ctrl-c isnt the best way to exit and exit , quit .. etc dont seem to do anything.

Torque Owner John Quigley
#9
04/21/2002 (4:46 pm)
g++ should be good about not leaving cruft in /tmp. You might just need a "make clean", or if that doesn't work, remove all of the "out.GCC2.DEBUG" directories.

As for console commands, I don't know of any list, but you can look for all of the "ConsoleFunctions" in the engine code. The command to quit is "quit();"

Torque Owner Nick Carnevalino
#10
04/21/2002 (5:34 pm)
well it's starting to look like a 3rd party prorgram problem .. i'm fine as long as i dont run pine ( text based e-mail reader ) somthing i tend to do every so often while i'm working on the linux machine ... dont know what it is able that program but if i run the torquedemo i'm fine and stop start it as much as i want .. if i run pine then pine gives me a segmentation fault the first time i try but is fine from then on .. but once i run pine i have to re-build torque or torque gives that same segemntion fault.

as far as i know all pine does is parce out the text mail file and sort it into messages/folders ... dont know just whatelse that effects .. maby some std in/out routines in the kernel but that in no way should effect other programs. Maby torque needs a lock file or somthing else that pine changes in the ~/ dir?

anyway for now as long as i dont run pine i seem to be ok so i'm not going to push it right now:-)

Torque Owner John Quigley
#11
04/21/2002 (5:59 pm)
OK, the weirdness bar has climbed even higher. I don't get why pine could cause a problem, it accesses the ~/. directory but lots of programs do that. It might a problem with the dynamic linker. Hmmm....recently there was a glibc update via up2date. I just pulled it down myself. Wonder if you need to reboot after such updates? Not sure.

Oh well, as long as you have a workaround. It doesn't sound like a bug in torque itself.