Question

Defined in h file:

char *copiedText;

Defined in c file:

char *copyText(char *text){
  if (copiedText != 0) free(copiedText);
  copiedText = (char *)calloc((strlen(text) + 1) * sizeof(char), sizeof(char));
  strcpy(copiedText, text);
  return copiedText;
}

First off, this isn’t about how to copy text, I just chose that as an example. My question is about the free() before calloc().

First question I hear you ask is why not just free() at the appropriate time – i.e. when copiedText is no longer needed?

Long story short, I’m making part of a program and I can not trust the users of my function to properly free() copiedText, so I want to contain as much code inside my function as possible. All they do is include h and c file and call the function copyText.

I am aware there will be a (minor) memory leak between last call to copyText and program termination, but this is a tradeoff that I am willing to accept, because I only malloc() small amounts of data.

The question is simply, will the free() actually free up the memory allocated by my calloc() when coded like this?

Answer 1

The flip side is that now:

1. the pointer returned by copyText() is only valid until the next call to copyText() — this is not at all obvious to the caller and is probably worth documenting as prominently as you can;
2. the function is not thread-safe.

A better approach might be to require an explicit free() (perhaps wrapping it in a function called freeText(), as a counterpart to copyText()).

Another alternative is for the caller to provide their own buffer along with its size (so that you can avoid buffer overruns).