Initialization

A declaration of an object may provide its initial value through the process known as initialization.

For each declarator, the initializer, if not omitted, may be one of the following:

where initializer-list is a non-empty comma-separated list of initializers (with an optional trailing comma), where each initializer has one of three possible forms:

where designator-list is a list of either array designators of the form [ constant-expression ] or struct/union member designators of the form . identifier; see array initialization and struct initialization.

Explanation

The initializer specifies the initial value stored in an object.

Explicit initialization

If an initializer is provided, see.

• scalar initialization for the initialization of scalar types
• array initialization for the initialization of array types
• struct initialization for the initialization of struct and union types.

Implicit initialization

If an initializer is not provided:

• objects with automatic storage duration are initialized to indeterminate values (which may be trap representations)
• objects with static and thread-local storage duration are initialized as follows

• pointers are initialized to null pointer values of their types
• objects of integral types are initialized to unsigned zero
• objects of floating types are initialized to positive zero
• members of arrays, structs, and unions are initialized as described above, recursively, plus all padding bits are initialized to zero

(on platforms where null pointers and floating zeroes have all-bit-zero representations, this form of initialization for statics is normally implemented by allocating them in the .bss section of the program image)

Notes

When initializing an object of static or thread-local storage duration, every expression in the initializer must be a constant expression or string literal.

Initializers cannot be used in declarations of objects of incomplete type, VLAs, and block-scope objects with linkage.

The initial values of function parameters are established as if by assignment from the arguments of a function call, rather than by initialization (until the post-C11 defect report DR 427, which changes the wording to use initialization).

If an indeterminate value is used as an argument to any standard library call, the behavior is undefined. Otherwise, the result of any expression involving indeterminate values is an indeterminate value (e.g. int n; may not compare equal to itself and it may appear to change its value on subsequent reads).

Example

#include <stdlib.h>
int a[2]; // initializes a to {0, 0}
int main(void)
{
    int i;          // initializes i to an indeterminate value
    static int j;   // initializes j to 0
    int k = 1;      // initializes k to 1
 
    // initializes int x[3] to 1,3,5
    // initializes int* p to &x[0]
    int x[] = { 1, 3, 5 }, *p = x;
 
    // initializes w (an array of two structs) to
    // { { {1,0,0}, 0}, { {2,0,0}, 0} }
    struct {int a[3], b;} w[] = {[0].a = {1}, [1].a[0] = 2};
 
    // function call expression can be used for a local variable
    char* ptr = malloc(10);
    free(ptr);
 
//  Error: objects with static storage duration require constant initializers
//  static char* ptr = malloc(10);
 
//  Error: VLA cannot be initialized
//  int vla[n] = {0};
}

References

• C11 standard (ISO/IEC 9899:2011):
• 6.7.9 Initialization (p: 139-144)
• C99 standard (ISO/IEC 9899:1999):
• 6.7.8 Initialization (p: 125-130)
• C89/C90 standard (ISO/IEC 9899:1990):
• 6.5.7 Initialization

See also