A pointer is declared much like any other variable, except an asterisk (*
) is placed between the type and the name of the variable to denote it is a pointer.
int *pointer; /* inside a function, pointer is uninitialized and doesn't point to any valid object yet */
To declare two pointer variables of the same type, in the same declaration, use the asterisk symbol before each identifier. For example,
int *iptr1, *iptr2;
int *iptr3, iptr4; /* iptr3 is a pointer variable, whereas iptr4 is misnamed and is an int */
The address-of or reference operator denoted by an ampersand (&
) gives the address of a given variable which can be placed in a pointer of appropriate type.
int value = 1;
pointer = &value;
The indirection or dereference operator denoted by an asterisk (*
) gets the contents of an object pointed to by a pointer.
printf("Value of pointed to integer: %d\n", *pointer);
/* Value of pointed to integer: 1 */
If the pointer points to a structure or union type then you can dereference it and access its members directly using the ->
operator:
SomeStruct *s = &someObject;
s->someMember = 5; /* Equivalent to (*s).someMember = 5 */
In C, a pointer is a distinct value type which can be reassigned and otherwise is treated as a variable in its own right. For example the following example prints the value of the pointer (variable) itself.
printf("Value of the pointer itself: %p\n", (void *)pointer);
/* Value of the pointer itself: 0x7ffcd41b06e4 */
/* This address will be different each time the program is executed */
Because a pointer is a mutable variable, it is possible for it to not point to a valid object, either by being set to null
pointer = 0; /* or alternatively */
pointer = NULL;
or simply by containing an arbitrary bit pattern that isn't a valid address. The latter is a very bad situation, because it cannot be tested before the pointer is being dereferenced, there is only a test for the case a pointer is null:
if (!pointer) exit(EXIT_FAILURE);
A pointer may only be dereferenced if it points to a valid object, otherwise the behavior is undefined. Many modern implementations may help you by raising some kind of error such as a segmentation fault and terminate execution, but others may just leave your program in an invalid state.
The value returned by the dereference operator is a mutable alias to the original variable, so it can be changed, modifying the original variable.
*pointer += 1;
printf("Value of pointed to variable after change: %d\n", *pointer);
/* Value of pointed to variable after change: 2 */
Pointers are also re-assignable. This means that a pointer pointing to an object can later be used to point to another object of the same type.
int value2 = 10;
pointer = &value2;
printf("Value from pointer: %d\n", *pointer);
/* Value from pointer: 10 */
Like any other variable, pointers have a specific type. You can't
assign the address of a short int
to a pointer to a long int
, for
instance. Such behavior is referred to as type punning and is forbidden in C, though there are a few exceptions.
Although pointer must be of a specific type, the memory allocated for each type of pointer is equal to the memory used by the environment to store addresses, rather than the size of the type that is pointed to.
#include <stdio.h>
int main(void) {
printf("Size of int pointer: %zu\n", sizeof (int*)); /* size 4 bytes */
printf("Size of int variable: %zu\n", sizeof (int)); /* size 4 bytes */
printf("Size of char pointer: %zu\n", sizeof (char*)); /* size 4 bytes */
printf("Size of char variable: %zu\n", sizeof (char)); /* size 1 bytes */
printf("Size of short pointer: %zu\n", sizeof (short*)); /* size 4 bytes */
printf("Size of short variable: %zu\n", sizeof (short)); /* size 2 bytes */
return 0;
}
(NB: if you are using Microsoft Visual Studio, which does not support the C99 or C11 standards, you must use %Iu
1 instead of %zu
in the above sample.)
Note that the results above can vary from environment to environment in numbers but all environments would show equal sizes for different types of pointer.
Extract based on information from Cardiff University C Pointers Introduction
Pointers and arrays are intimately connected in C. Arrays in C are always held in contiguous locations in memory. Pointer arithmetic is always scaled by the size of the item pointed to. So if we have an array of three doubles, and a pointer to the base, *ptr
refers to the first double, *(ptr + 1)
to the second, *(ptr + 2)
to the third. A more convenient notation is to use array notation []
.
double point[3] = {0.0, 1.0, 2.0};
double *ptr = point;
/* prints x 0.0, y 1.0 z 2.0 */
printf("x %f y %f z %f\n", ptr[0], ptr[1], ptr[2]);
So essentially ptr and the array name are interchangeable. This rule also means that an array decays to a pointer when passed to a subroutine.
double point[3] = {0.0, 1.0, 2.0};
printf("length of point is %s\n", length(point));
/* get the distance of a 3D point from the origin */
double length(double *pt)
{
return sqrt(pt[0] * pt[0] + pt[1] * pt[1] + pt[2] * pt[2])
}
A pointer may point to any element in an array, or to the element beyond the last element. It is however an error to set a pointer to any other value, including the element before the array. (The reason is that on segmented architectures the address before the first element may cross a segment boundary, the compiler ensures that does not happen for the last element plus one).
Footnote 1: Microsoft format information can be found via printf()
and format specification syntax.