C code stops running

Question

I have the following C code which gives an error:

Program stopped at 0x4019b3.
It stopped with signal SIGSEGV, Segmentation fault.

when debugging.

Here is the code:

#include <stdio.h>
#include <complex.h>
#include <stdlib.h>
#include <time.h>
int main()
{
    clock_t begin, end;
    double time_spent;

    begin = clock();

    int n  = 100;int i; int j;
    int N  = 64;int r;
    double complex (s)[4] = {-1-1*I, -1+1*I, 1-1*I, 1+1*I};
    double complex symbol[n][N];
    for (i=0; i<n; i++){
        for (j=0; j<N; j++){
            r = rand() % 4;
            symbol[i][j]=s[r];  
        }
        // Now add pilots:
        symbol[i][11] = 1;
        symbol[i][22] = 1;
        symbol[i][33] = 1;
        symbol[i][44] = 1;
    } 
    end = clock();
    time_spent = (double)(end - begin) / CLOCKS_PER_SEC;
    return 0;
}

Any idea what's wrong?

EDIT:

Now I can put it all together after these valuable discussions. Here is the working code with timing and memory allocation and every thing:

#include <stdio.h>
#include <complex.h>
#include <stdlib.h>
#include <time.h>
int main()
{
    clock_t begin, end;
    double time_spent;
    begin = clock();

    int n  = 100000; int i; int j;
    int N  = 64;int r;
    double complex (s)[4] = {-1-1*I, -1+1*I, 1-1*I, 1+1*I};
    double complex (*symbol)[N] = malloc(n * sizeof *symbol);
    for (i=0; i<n; i++){
        for (j=0; j<N; j++){
            r = rand() % 4;
            symbol[i][j]=s[r];  
        }
        // Now add pilots:
        symbol[i][11] = 1;
        symbol[i][22] = 1;
        symbol[i][33] = 1;
        symbol[i][44] = 1;
    } 
    end = clock();
    time_spent = (double)(end - begin) / CLOCKS_PER_SEC;
    printf("%3.7f\n",time_spent);
    return 0;
}

What have you done -- if anything - to try and figure out this issue? — Nik Bougalis, Sep 29 '14 at 02:28
This line is wrong: `double complex (*symbol)[n][N] = malloc(n * sizeof *symbol);`. Copy more carefully out of R Sahu's answer. Also you have sprouted bogus `*` everywhere. It should be `symbol[i][j]`, not `*symbol[i][j]` and the same for all the pilots. — M.M, Sep 29 '14 at 02:58
If you don't understand something then look up how it works, don't just guess something and hit compile and then get puzzled when there is a segfault — M.M, Sep 29 '14 at 03:01
Ok! I used `printf("%3.7f\n",time_spent);` and got `0.0010000` sounds reasonable? — AboAmmar, Sep 29 '14 at 03:31

R Sahu · Accepted Answer · 2014-09-29T02:51:22.560

3

Memory needed to hold the variable declared by the line

double complex symbol[100000][64];

is too much for the stack.

Even a simple program like below runs into Segmentation fault when run an a 64-bit machine.

#include <stdio.h>
#include <complex.h>

void foo()
{
   double complex symbol[100000][64];
   printf("%zu\n", sizeof(symbol));
}

int main(int argc, char** argv)
{
   foo();
   return 0;
}

Consider allocating that memory from the heap, for example:

double complex (*symbol)[N] = malloc(n * sizeof *symbol);

The other problem is that in the loops:

for (i=0; i<n; i++){
    for (j=0; i<N; j++){  // Problem line
        r = rand() % 4;
        symbol[i][j]=s[r];  
    }

You are accessing out of bounds memory. The problem line should be changed to:

    for (j=0; j<N; j++){
              ^^ Use j not i

edited Sep 29 '14 at 02:51

answered Sep 29 '14 at 02:31

R Sahu

204,454
14
159
270

2

Where there's one problem, there's usually more than one… – Jonathan Leffler Sep 29 '14 at 02:34
@R Sahu It is not memory. I try `symbol[10][64]` and get the same error. But the debugger points to the line `symbol[i][j]=s[r];`. – AboAmmar Sep 29 '14 at 02:41
1

@AboAmmar you are accessing out of bounds of the `symbol` array as suggested by MitchWheat's answer. You need to fix *both* of these problems for your code to work. – M.M Sep 29 '14 at 02:49
@MattMcNabb I already corrected `i` to `j`. Same error on line `*symbol[i][j]=s[r];` – AboAmmar Sep 29 '14 at 02:57
@AboAmmar your latest edit screws everything up totally – M.M Sep 29 '14 at 02:58
Your latest edit shows `r = rand()/4;`. Is that your real code or is that a typ0? It should be `r = rand()%4;`. – R Sahu Sep 29 '14 at 02:59
@R Sahu yes it is `r = rand()/4;` – AboAmmar Sep 29 '14 at 03:00
@AboAmmar, That would not keep `r` between `0` and `3`. That's a problem. – R Sahu Sep 29 '14 at 03:02
@AboAmmar it should be `r = rand() % 4;` – M.M Sep 29 '14 at 03:02
@MattMcNabb I'm 100% sure it is in this line `r = rand()/4;`. because when I replace it with a constant I have no error. – AboAmmar Sep 29 '14 at 03:08
@AboAmmar you've been told several times now that that line is a mistake , try reading what people are saying – M.M Sep 29 '14 at 03:14
@MattMcNabb I used it both ways with the same error. – AboAmmar Sep 29 '14 at 03:17
@AboAmmar think about what you are writing, instead of guessing – M.M Sep 29 '14 at 03:18
Yes it is now fixed using `r = rand() % 4`. – AboAmmar Sep 29 '14 at 03:20
But how to get the time? – AboAmmar Sep 29 '14 at 03:20
@AboAmmar, With a lot of back and forth, you got over the segmentation fault error. Try to take it from there. – R Sahu Sep 29 '14 at 03:23

Mitch Wheat · Answer 2 · 2014-09-29T02:33:28.647

1

You have a copy and paste error in your second loop test:

for (j=0; j<N; j++){
          ^

It should be j not i

[Also, unrelated but you should not use modulus % on the result of rand() because the low bits are not as random as the high bits. Use division instead.]

[Another answer points out that you may also be exhausting your stack, although I would expect a different error. Worth checking though.]

edited Sep 29 '14 at 02:33

answered Sep 29 '14 at 02:28

Mitch Wheat

295,962
43
465
541

Re. the randomness issue, that would depend on which PRNG was in use – M.M Sep 29 '14 at 02:32
@MitchWheat The same error and, BTW, when I run it a window with `ProgName.exe has stopped responding` appears above the black `.exe` output window. – AboAmmar Sep 29 '14 at 02:33
@Matt: stdlib.h, rand() – Mitch Wheat Sep 29 '14 at 02:37
@MitchWheat `rand()` is a standard C function, the standard does not specify that lower order bits must have lower randomness than higher order bits. The OP's implementation might have one in which lower order bits do not have lesser randomness than higher order bits. – M.M Sep 29 '14 at 02:38
I assume you mean LCG. Glibc [defaults to LFSR](http://stackoverflow.com/questions/3932978/gcc-implementation-of-rand) so I question your statistic. It might be better to link to discussion of this topic instead of asserting. – M.M Sep 29 '14 at 02:45

C code stops running

2 Answers2