Problem writing int to binary file in C

Question

I need to write data to a binary file using C's I/O functions. The following code causes a runtime exception :

#include "stdio.h"

int main(int argc,char* argv[]) {
    FILE *fp = fopen("path_to_file.bin","wb");
    if(fp == NULL) {
     printf("error creating file");
     return -1;
    }
    int val = 4;
    fwrite((const void*)val,sizeof(int),1,fp);
    fclose(fp);
    return 0;
}

The code dies at the fwrite. Can you spot what I'm doing wrong? Apparently, I'm, trying to access data at 0x0000004 or something like that.

Thanks !

Answer 1

 fwrite((const void*)val,sizeof(int),1,fp);

should be:

 fwrite((const void*) & val,sizeof(int),1,fp);

BTW, if you don't use the cast, you will get a sensible error message. Casts tend to be used by C (and C++) programmers far more often than they should be - a good rule of thumb is "if it needs a cast, it's probably wrong".

Answer 2

Adding to Neil's answer: this works when you are reading and writing the file on the same platform. Things can become weird if you are reading/writing across platforms with different endianness.

Answer 3

I think Neil's answer can be improved upon. I realize it's already accepted, so this is just to show some contrast (which is why I didn't just edit his).

fwrite(&val, sizeof val, 1, fp);

Two improvements:

• No pointer casting, since it's not necessary in C and can hide errors.
• Use sizeof directly on the object, since that is what you're passing a pointer to. Makes a lot of sense to me, and is safer than repeating yourself and using the type name.

Answer 4

#include "stdio.h"

int main(int argc,char* argv[]) {
    FILE *fp = fopen("path_to_file.bin","wb");
    if(fp == NULL) {
        printf("error creating file");
        return -1;
    }
    int val = 4;
    fwrite((const void*)val,sizeof(int),1,fp);

You should supply an address not integer itself.

Additionaly you should not use the integer in such way:

1. It may differ in endianess on different computers (as mentioned)
2. It may differ in size. On really old computers it may be 1 or 2 bytes. On most modern it will be 4 but it may be 8 as well (some 64-bit computers). On some strange architectures it may be even 36 bits. int32_t val = 4; fwrite((const void *)val, 4, 1, fp) should solve the problem.

You may think that your software will never need to be ported. Well - many designers (software and hardware) made similar assumptions. Sometimes it is too costly to not make them - but in this case it is just a matter of few additional checks.

    fclose(fp);
    return 0;
}