Thread: segmentation faults

Hi guys,
I was running a simple code on Fedora 11. it has this error " segmentation fault". The code is too simple to have any mistakes........ Can anyone help me with this? What is this "segmentation fault".

Thank you very much.

Code:

#include "stdio.h" 
int main()
{
int T[1001][4001]; 
int i=0;
int j=0; 
for(j=0;j<1001;j++)
{ 
for(i=0;i<4001;i++)
{ 
T[j][i]=0; 
}
} 

printf("Thank you!\n"); 

return 0;
}

I saved this file as mainapp.cpp.

command line:

g++ mainapp.cpp -o test
./test

then it says segmentation fault........

The probability of you being allowed to have four million ints on the stack is zero. If you need four million ints, you will have to manage that memory yourself (with malloc/free/and friends).

Originally Posted by tabstop

The probability of you being allowed to have four million ints on the stack is zero. If you need four million ints, you will have to manage that memory yourself (with malloc/free/and friends).

Thanks for the help. Could you tell me how to change the size of the stack. I know with VC++, it is easy to resize the stack. But in Linux, how to do it?

Originally Posted by Adak

I don't program in linux, but I suspect that you're just asking for too much memory, on the stack.

That's a truck load of bytes there, partner!

Try it with 1,000 and see what happens.

Segmentation fault is an "Oh ........, my engine broke the crank", kind of error. Usually occurs when something in your program (like a pointer), refers to something outside of it's allocated memory space.

If you need more memory than you can get this way, try making it a global (before main()). If that still isn't enough (and it won't be enough for this amount), then you'll need to malloc() that memory from the heap. The memory must not only be available, in the quantity you're requesting, but it must also be available in one contiguous "chunk".

For your program code, I have two suggestions:

1) Use spaces instead of tabs - tabs don't always work with the forum software, just right. Two to four spaces, is ideal for each level of indentation.

2) Use indentation, always. Don't leave home without it, buddy!

Good on ya, for using code tags around your program - you're ahead of the curve, already.

Thanks for the tips....... But I still do not know how to resize the stack? I know how to do it in VC++. but in Linux,, I dont know how.......... Could you tell me how to do it?

You should try this before running GDB

ulimit -c unlimited

It might increase the size of stack....

Originally Posted by subhashish1213

You should try this before running GDB

ulimit -c unlimited

It might increase the size of stack....

Nope, that will only increase the allowable size of a core file. What he wants to do, (assuming he insists on ignoring Tabstop's advice about using dynamic memory allocation) is to issue the ulimit -s command

At the OP, you can read about it in your ulimit man page, but if you do

Code:

ulimit -a

you will get a listing of your shell resource limits. There are soft and hard resource limits, and the foregoing command will show you (if any) the soft limits. By doing

Code:

ulimit -Ha

you will get the same listing of all (-a) limits, but now they will be displayed with the hard (-H) limits. In the case of your Fedora install, I am willing to bet that the soft stack size will be 10240 kilobytes, and the hard limit will be unlimited. To change the size of the stack, just do

Code:

ulimit -s <new size in kb>

So then, if you wanted the stack to be 20500 kb you would do:

Code:

ulimit -s 20500

If you wanted that as your hard limit, you would do:

Code:

ulimit -Hs 20500

If your original stack hard limit is unlimited (and I would guess it is), then you can get to that limit by doing:

Code:

ulimit -s unlimited

A couple of things to remember though:

1) The change you make is not persistent across logins. Once you logout of that shell, the change is gone.

2) Once you set the upper limit, you will not be allowed past that. If you set it too low, you will need to log in to a new shell.

Finally, I put all of this up here for information purposes, but I really think it is best, as per Tabstop's advice, to use dynamic memory allocation.