Objects and alignment

From cppreference.com
< c‎ | language

C programs create, destroy, access, and manipulate objects.

An object, in C, is region of data storage in the execution environment, the contents of which can represent values (a value is the meaning of the contents of an object, when interpreted as having a specific type).

Every object has

  • size (can be determined with sizeof)
  • alignment requirement (can be determined by alignof) (since C11)
  • storage duration (automatic, static, allocated, thread-local)
  • lifetime (equal to storage duration or temporary)
  • effective type (see below)
  • value (which may be indeterminate)
  • optionally, an identifier that denotes this object

Objects are created by declarations, allocation functions, string literals, compound literals, and by non-lvalue expressions that return structures or unions with array members.

Object representation

Except for bit fields, objects are composed of contiguous sequences of one or more bytes, each consisting of CHAR_BIT bits, and can be copied with memcpy into an object of type unsigned char[n], where n is the size of the object. The contents of the resulting array are known as object representation.

If two objects have the same object representation, they compare equal (except if they are floating-point NaNs). The opposite is not true: two objects that compare equal may have different object representations because not every bit of the object representation needs to participate in the value. Such bits may be used for padding to satisfy alignment requirement, for parity checks, to indicate trap representations, etc.

If an object representation does not represent any value of the object type, it is known as trap representation. Accessing a trap representation in any way other than reading it through an lvalue expression of character type is undefined behavior. The value of a structure or union is never a trap representation even if any particular member is one.

For the objects of type char, signed char, and unsigned char, every bit of the object representation is required to participate in the value representation and each possible bit pattern represents a distinct value (no padding, trap bits, or multiple representations allowed).

Effective type

Every object has an effective type, which determines which lvalue accesses are valid and which violate the strict aliasing rules.

If the object was created by a declaration, the declared type of that object is the object's effective type.

If the object was created by an allocation function (including realloc), it has no declared type. Such object acquires an effective type as follows:

  • The first write to that object through an lvalue that has a type other than character type, at which time the type of that lvalue becomes this object's effective type for that write and all subsequent reads.
  • memcpy or memmove copy another object into that object, at which time the effective type of the source object (if it had one) becomes the effective type of this object for that write and all subsequent reads.
  • Any other access to the object with no declared type, the effective type is the type of the lvalue used for the access.

Strict aliasing

Given an object with effective type T1, using an lvalue expression (typically, dereferencing a pointer) of a different type T2 is undefined behavior, unless:

  • T2 and T1 are compatible types.
  • T2 is cvr-qualified version of a type that is compatible with T1.
  • T2 is a signed or unsigned version of a type that is compatible with T1.
  • T2 is an aggregate type or union type type that includes one of the aforementioned types among its members (including, recursively, a member of a subaggregate or contained union).
  • T2 is a character type (char, signed char, or unsigned char).
int i = 7;
char* pc = (char*)(&i);
if(pc[0] == '\x7') // aliasing through char is ok
    puts("This system is little-endian");
else
    puts("This system is big-endian");
 
float* pf = (float*)(&i);
float d = *pf; // UB: float lvalue *p cannot be used to access int

Note that type-punning may also be performed through the inactive member of a union.

Alignment

Every complete object type has a property called alignment requirement, which is an integer value of type size_t representing the number of bytes between successive addresses at which objects of this type can be allocated. The valid alignment values are non-negative integral powers of two.

The alignment requirement of a type can be queried with alignof.

(since C11)

In order to satisfy alignment requirements of all members of a struct, padding may be inserted after some of its members.

#include <stdio.h>
#include <stdalign.h>
 
// objects of struct S can be allocated at any address
// because both S.a and S.b can be allocated at any address
struct S {
  char a; // size: 1, alignment: 1
  char b; // size: 1, alignment: 1
}; // size: 2, alignment: 1
 
// objects of struct X must be allocated at 4-byte boundaries
// because X.n must be allocated at 4-byte boundaries
// because int's alignment requirement is (usually) 4
struct X {
  int n;  // size: 4, alignment: 4
  char c; // size: 1, alignment: 1
  // three bytes padding
}; // size: 8, alignment: 4
 
int main(void)
{
    printf("sizeof(struct S) = %zu\n", sizeof(struct S));
    printf("alignof(struct S) = %zu\n", alignof(struct S));
    printf("sizeof(struct X) = %zu\n", sizeof(struct X));
    printf("alignof(struct X) = %zu\n", alignof(struct X));
}

Possible output:

sizeof(struct S) = 2
alignof(struct S) = 1
sizeof(struct X) = 8
alignof(struct X) = 4

Each object type imposes its alignment requirement on every object of that type. The strictest (largest) fundamental alignment of any type is the alignment of max_align_t. The weakest (smallest) alignment is the alignment of the types char, signed char, and unsigned char, and equals 1.

If an object's alignment is made stricter (larger) than max_align_t using alignas, it has extended alignment requirement. A struct or union type whose member has extended alignment is an over-aligned type. It is implementation-defined if over-aligned types are supported, and their support may be different in each kind of storage duration. (since C11)

References

  • C11 standard (ISO/IEC 9899:2011):
  • 3.15 object (p: 6)
  • 6.2.6 Representations of types (p: 44-46)
  • 6.5/6-7 Expressions (p: 77)
  • 6.2.8 Alignment of objects (p: 48-49)
  • C99 standard (ISO/IEC 9899:1999):
  • 3.2 alignment (p: 3)
  • 3.14 object (p: 5)
  • 6.2.6 Representations of types (p: 37-39)
  • 6.5/6-7 Expressions (p: 67-68)
  • C89/C90 standard (ISO/IEC 9899:1990):
  • 1.6 Definitions of terms

See also