using realloc() leads to segmentation fault after passing specific memory value

Question

I encountered a problem which after expanding memory with realloc(), Segmentation fault reached when it try to reach specific memory number. 135,126th char address to be exact.

But, that mark never a problem if I allocate memory with malloc() bigger in the first place.

Here is my experiment with C that run on linux AMD64 compiled with gcc:

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#define SECT_SIZE 25
#define SECT_EXTEND_SIZE 100
#define MAX_SECT_TEST 18000

void extend_memory(char *c_memory, int *i_current_sect_size){

    char *c_temp = realloc(c_memory, sizeof(char) * SECT_SIZE * (*i_current_sect_size + SECT_EXTEND_SIZE));
    if(c_temp == NULL){
        printf("realloc failed upon %d + %d\n", *i_current_sect_size, SECT_EXTEND_SIZE);
        exit(EXIT_FAILURE);
    }
    else{
        c_memory = c_temp;
        *i_current_sect_size += SECT_EXTEND_SIZE;
        printf("realloc success\n");
    }
}

int main(int argc, char *argv[]){

    //INITIALIZATION
    int i_current_sect_size;

    char *c_memory = (char *) malloc( sizeof(char) * SECT_SIZE * SECT_EXTEND_SIZE);
    if(c_memory == NULL){
        printf("c_memory malloc failed\n");
        exit(EXIT_FAILURE);
    } 
    else
        i_current_sect_size = SECT_EXTEND_SIZE;

    memcpy(c_memory, "123456789012345", SECT_SIZE); //fill up 0th sector.
    //INITIALIZATION FINISHED


    for(int a = 0; a < MAX_SECT_TEST; a++){ //fill dummy data foreach sectors 
        if(a + 1 == i_current_sect_size)
            extend_memory(c_memory, &i_current_sect_size);

        memcpy(&c_memory[SECT_SIZE * a], &c_memory[SECT_SIZE * (a+1)], SECT_SIZE);
        printf("success %d\n", a);
    }
    return EXIT_SUCCESS;
}

In simple words, this program would repeatedly copy previous SECT value into next SECT as counter a increase. Upon reaching maximum allocated address, it would expand memory allocation using realloc() by SECT_EXTEND_SIZE and continue doing its job. This process done when MAX_SECT_TEST reached.

I ran the program with SECT_SIZE equal to 9, 16, 25; All variations seems to fall into Segmentation Fault as it try to copy into SECT 15015, 8446, 5405 respectively.

9 x 15015 = 135,135

16 x 8446 = 135,136

25 x 5405 = 135,125

Why (135,125-135,136)th char become a limit? I have a 8GB memory by the way. I'm pretty sure the actual limit should way bigger than that.

memcpy(c_memory, "123456789012345", SECT_SIZE); reads out of bounds (SECT_SIZE is 25) — M.M
– M.M, Commented Mar 11, 2016 at 22:15
Your extend_memory is essentially a wrapper for realloc. Why did you design its interface to be different from realloc? Why aren't you returning the new pointer value, like realloc does? — AnT stands with Russia
– AnT stands with Russia, Commented Mar 11, 2016 at 22:25
@M.M thanks dude. realloc(*c_memory, sizeof(char) * SECT_SIZE * (*i_current_sect_size + SECT_EXTEND_SIZE)); need to changed as well besides two lines you corrected. — Shaziq
– Shaziq, Commented Mar 11, 2016 at 22:26
@AnT To avoid realloc() failure, I guess, as suggested in other realloc() related questions. — Shaziq
– Shaziq, Commented Mar 11, 2016 at 22:34
@Shaziq: But you have already properly taken care of any possible realloc failures inside extend_memory (which is probably one of the reasons your extend_memory even exists). Now, once you've taken care of this issue, why wouldn't you keep the external interface unchanged? I mean, one big problem with the code is that you do not properly return the result to the calling code. You can fix it by going the ** route (as suggested in the answers), or you can simply literally return it from the function. The latter looks more elegant to me. — AnT stands with Russia
– AnT stands with Russia, Commented Mar 11, 2016 at 22:54

ensc · Accepted Answer · 2016-03-11 22:03:27Z

3

void extend_memory(char *c_memory, int *i_current_sect_size){
        ...
        c_memory = c_temp;

is effectively a noop and caller will not see the changed c_memory.

You will need to use an argument like char **c_memory and work with *c_memory within the function.

answered Mar 11, 2016 at 22:03

ensc

7,06220 silver badges25 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

AnT stands with Russia Over a year ago

Either that, or make extend_memory return the new pointer, just like realloc does.

ensc Over a year ago

true; but because size is also adjusted and returned in a by-reference argument it might be more consistent to keep it as a parameter. Alternatively, a struct buffer { void *mem; size_t len }; can be declared and used.

John Bode · Accepted Answer · 2016-03-11 22:19:44Z

1

If you want a function to update a pointer value, you must pass a pointer to the pointer:

extend_memory(&c_memory, &i_current_sect_size);

and update your function as

void extend_memory(char **c_memory, int *i_current_sect_size){
...
  char *c_temp = realloc(*c_memory, sizeof(char) * SECT_SIZE * (*i_current_sect_size + SECT_EXTEND_SIZE));
...
  *c_memory = c_temp;

answered Mar 11, 2016 at 22:19

John Bode

125k19 gold badges130 silver badges211 bronze badges

Collectives™ on Stack Overflow

using realloc() leads to segmentation fault after passing specific memory value

2 Answers 2

2 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related