Thread: how the malloc function work in C

I've been trying to understand how the malloc function in C dynamically allocates memory, and I'm not entirely clear on it. I read some theory but still have some questions.

Let's look at this line of code as an example:

Code:

int *ptr = (int *)malloc(sizeof(int));

I understand that ptr is declared as a pointer to an integer. What's puzzling me is the statement that "malloc returns a pointer to the first byte of the allocated memory."


If I get it correctly, ptr should hold the memory address where integer variables can be stored. But what does it mean by "malloc returns a pointer to the first byte of the allocated memory"? Does this imply that ptr is holding another pointer that points to the actual memory?

Originally Posted by Salem

All pointers point to the first byte of whatever it is they're pointing to.

Thank you salem
What's puzzling me is the statement that "malloc returns a pointer to the first byte of the allocated memory."

So, a function's return type can vary depending on the specific requirements . It can return values, pointer


I believe that when the malloc function is executed, it returns a void pointer (void*), and the value it returns is, in fact, the memory address that the pointer ptr holds.

I'd like to confirm if my understanding is correct.

Thank you salem
What's puzzling me is the statement that "malloc returns a pointer to the first byte of the allocated memory."

Why does this puzzle you? A pointer is a memory address. By almost-universal convention, pointers contain the address of the first byte of the data structure being pointed to. There are exceptions to this, however, including malloc()! Some data structures are stored with a "prefix" that occurs before the address that is pointed to. Generally,
this is done to avoid having the data overwritten, or to allow the pointer to serve multiple purposes. Examples:

1. In C, a common idiom for dynamic arrays (what C++ calls a vector) is to store length and capacity in a prefix struct that appears at the beginning of the allocated storage, then store the array. The returned pointer always references the start of the array, and some pointer math has to be done to look up the prefix struct.

2. In C++ a common implementation of multiple inheritance is implemented by partitioning object memory into distinct parts for each parent class. Since inheritance can involve virtual methods, each distinct part stores a separate vtable as well, making it possible to pass a pointer into the middle of the child object to serve as the address for different parent classes.

So, a function's return type can vary depending on the specific requirements . It can return values, pointer

In C, a function's return type is set in stone when the function is declared. The return type can never be other than what it was declared with. You can certainly change the declaration as a developer, and you have the option of returning a union or a struct containing multiple data items. But once the code is compiled, the return type is fixed.

I believe that when the malloc function is executed, it returns a void pointer (void*), and the value it returns is, in fact, the memory address that the pointer ptr holds.

Technically, there are quibbles about whether NULL is a pointer or not. But you are substantially correct.

Originally Posted by aghast

Technically, there are quibbles about whether NULL is a pointer or not. But you are substantially correct.

NULL the constant may be either a pointer or just a plain 0. So an implementation is allowed to do either of these:

Code:

#define NULL 0

or

Code:

#define NULL ((void *)0)

(or a variation of one of those).

A null pointer (one that has been set to NULL or 0) is still a pointer, but it just happens to point to nothing.

Note: yes, there are some architectures where a null pointer is not all zero bits, but even on those architectures, NULL is still defined as 0. C takes care of converting a 0 constant to whatever bit pattern the architecture uses for null pointers. In the vast majority of cases you don't have even to worry about it. See Question 5.2 How do I get a null pointer in my programs?, Question 5.4 What is NULL and how is it defined?, and Question 5.5 How should NULL be defined on a machine which uses a nonzero bit pattern as the internal representation of a null pointer?. C++ also has a nullptr keyword which is basically equivalent to NULL.