freeing mem

i have two functions:

Code:

MY_TYPE *dostuff()
{
      MY_TYPE *my_type;
      ...
      return my_type;
}

void other()
{
      MY_TYPE *my_type;
      ...
      my_type = dostuff();
      ...
      free(my_type);
}

- both pointers point at the same data.
- only the one free is needed to free all memory.

is this true?

If I understand you correctly, yes, its true. (I presume you're malloc()'ing memory in the middle of dostuff() somewhere?).

For each chunk of memory malloc()'d, you must issue one free(). You can store the address of that memory is as many pointers as you like, but you must only free() one of them.

yes i do mean that.

its actually a MYSQL_RES type i'm using.

you don't have to malloc it, but you have to call
mysql_free_result()!

thanks!

hammer, i'm also having some reallocing problems - it also has to do with freeing mem!

i had a malloc:

Code:

int *sites; //in struct

my_pd->sites = malloc(sizeof(int) * my_pd->max_site_number);

during testing, i realised i was leaking mem and found that this line was getting called many times in a loop.

question 1 : will this allocate more and more mem to sites (i.e. leak mem)?

thinking it would i changed the code to:

Code:

realloc(my_pd->sites,sizeof(int) * my_pd->max_site_number);

if(my_pd->sites != NULL){free(my_pd->sites);} //when done

question 2: is this correct? why do i now get a random seg fault?

thanks!

It's a little difficult to tell without seeing more of the code... but here goes anyway...

>>leaking mem and found that this line was getting called many times in a loop.
If you are simply malloc() memory, storing the pointer, then repeating the process, then yes, you are probably throwing away pointers to malloc()'d memory, which is obviously bad.

>i changed the code to...
Is realloc() inside a loop? Is the value of max_site_number static? If yes to both, whats the point?

>realloc(my_pd->sites,....
When you first call realloc(), what's sites value? If it doesn't point anywhere in particular, it should be set to NULL. And I presume you are trapping the return code from realloc() somewhere?

If you're wanting dymanic array advice, have a read of this.

thanks hammer, i think i'm going mad,

am i right in saying then, that realloc should only be called after malloc (or calloc).

its ok to malloc and free the same variable many times in a loop?

by the way i'm in oxfordshire!

>am i right in saying then, that realloc should only be called after malloc (or calloc).
No, not correct! If you pass a NULL pointer to realloc() as its first arg, you simulate malloc(). That way you save yourself the bother of firt malloc()'ing then realloc()'ing.

>its ok to malloc and free the same variable many times in a loop?
Yes. Although the question would be why would you want to?

>by the way i'm in oxfordshire!
Ahh.. another UK bod

thanks hammer,

i need to get a good c book really. i have more knowledge of oo (c++ & java).

can you recommend one for my level.

There are a few good and a few bad. Do a few searches on here and the web. Here's a good place to start:
http://www.accu.org/bookreviews/public/