Unsafe code

The core C# language, as defined in the preceding chapters, differs notably from C and C++ in its omission of pointers as a data type. Instead, C# provides references and the ability to create objects that are managed by a garbage collector. This design, coupled with other features, makes C# a much safer language than C or C++. In the core C# language it is simply not possible to have an uninitialized variable, a "dangling" pointer, or an expression that indexes an array beyond its bounds. Whole categories of bugs that routinely plague C and C++ programs are thus eliminated.

While practically every pointer type construct in C or C++ has a reference type counterpart in C#, nonetheless, there are situations where access to pointer types becomes a necessity. For example, interfacing with the underlying operating system, accessing a memory-mapped device, or implementing a time-critical algorithm may not be possible or practical without access to pointers. To address this need, C# provides the ability to write unsafe code.

In unsafe code it is possible to declare and operate on pointers, to perform conversions between pointers and integral types, to take the address of variables, and so forth. In a sense, writing unsafe code is much like writing C code within a C# program.

Unsafe code is in fact a "safe" feature from the perspective of both developers and users. Unsafe code must be clearly marked with the modifier unsafe, so developers can't possibly use unsafe features accidentally, and the execution engine works to ensure that unsafe code cannot be executed in an untrusted environment.

Unsafe contexts

The unsafe features of C# are available only in unsafe contexts. An unsafe context is introduced by including an unsafe modifier in the declaration of a type or member, or by employing an unsafe_statement:

  • A declaration of a class, struct, interface, or delegate may include an unsafe modifier, in which case the entire textual extent of that type declaration (including the body of the class, struct, or interface) is considered an unsafe context.
  • A declaration of a field, method, property, event, indexer, operator, instance constructor, destructor, or static constructor may include an unsafe modifier, in which case the entire textual extent of that member declaration is considered an unsafe context.
  • An unsafe_statement enables the use of an unsafe context within a block. The entire textual extent of the associated block is considered an unsafe context.

The associated grammar productions are shown below.

class_modifier_unsafe
    : 'unsafe'
    ;

struct_modifier_unsafe
    : 'unsafe'
    ;

interface_modifier_unsafe
    : 'unsafe'
    ;

delegate_modifier_unsafe
    : 'unsafe'
    ;

field_modifier_unsafe
    : 'unsafe'
    ;

method_modifier_unsafe
    : 'unsafe'
    ;

property_modifier_unsafe
    : 'unsafe'
    ;

event_modifier_unsafe
    : 'unsafe'
    ;

indexer_modifier_unsafe
    : 'unsafe'
    ;

operator_modifier_unsafe
    : 'unsafe'
    ;

constructor_modifier_unsafe
    : 'unsafe'
    ;

destructor_declaration_unsafe
    : attributes? 'extern'? 'unsafe'? '~' identifier '(' ')' destructor_body
    | attributes? 'unsafe'? 'extern'? '~' identifier '(' ')' destructor_body
    ;

static_constructor_modifiers_unsafe
    : 'extern'? 'unsafe'? 'static'
    | 'unsafe'? 'extern'? 'static'
    | 'extern'? 'static' 'unsafe'?
    | 'unsafe'? 'static' 'extern'?
    | 'static' 'extern'? 'unsafe'?
    | 'static' 'unsafe'? 'extern'?
    ;

embedded_statement_unsafe
    : unsafe_statement
    | fixed_statement
    ;

unsafe_statement
    : 'unsafe' block
    ;

In the example

public unsafe struct Node
{
    public int Value;
    public Node* Left;
    public Node* Right;
}

the unsafe modifier specified in the struct declaration causes the entire textual extent of the struct declaration to become an unsafe context. Thus, it is possible to declare the Left and Right fields to be of a pointer type. The example above could also be written

public struct Node
{
    public int Value;
    public unsafe Node* Left;
    public unsafe Node* Right;
}

Here, the unsafe modifiers in the field declarations cause those declarations to be considered unsafe contexts.

Other than establishing an unsafe context, thus permitting the use of pointer types, the unsafe modifier has no effect on a type or a member. In the example

public class A
{
    public unsafe virtual void F() {
        char* p;
        ...
    }
}

public class B: A
{
    public override void F() {
        base.F();
        ...
    }
}

the unsafe modifier on the F method in A simply causes the textual extent of F to become an unsafe context in which the unsafe features of the language can be used. In the override of F in B, there is no need to re-specify the unsafe modifier -- unless, of course, the F method in B itself needs access to unsafe features.

The situation is slightly different when a pointer type is part of the method's signature

public unsafe class A
{
    public virtual void F(char* p) {...}
}

public class B: A
{
    public unsafe override void F(char* p) {...}
}

Here, because F's signature includes a pointer type, it can only be written in an unsafe context. However, the unsafe context can be introduced by either making the entire class unsafe, as is the case in A, or by including an unsafe modifier in the method declaration, as is the case in B.

Pointer types

In an unsafe context, a type (Types) may be a pointer_type as well as a value_type or a reference_type. However, a pointer_type may also be used in a typeof expression (Anonymous object creation expressions) outside of an unsafe context as such usage is not unsafe.

type_unsafe
    : pointer_type
    ;

A pointer_type is written as an unmanaged_type or the keyword void, followed by a * token:

pointer_type
    : unmanaged_type '*'
    | 'void' '*'
    ;

unmanaged_type
    : type
    ;

The type specified before the * in a pointer type is called the referent type of the pointer type. It represents the type of the variable to which a value of the pointer type points.

Unlike references (values of reference types), pointers are not tracked by the garbage collector -- the garbage collector has no knowledge of pointers and the data to which they point. For this reason a pointer is not permitted to point to a reference or to a struct that contains references, and the referent type of a pointer must be an unmanaged_type.

An unmanaged_type is any type that isn't a reference_type or constructed type, and doesn't contain reference_type or constructed type fields at any level of nesting. In other words, an unmanaged_type is one of the following:

  • sbyte, byte, short, ushort, int, uint, long, ulong, char, float, double, decimal, or bool.
  • Any enum_type.
  • Any pointer_type.
  • Any user-defined struct_type that is not a constructed type and contains fields of unmanaged_types only.

The intuitive rule for mixing of pointers and references is that referents of references (objects) are permitted to contain pointers, but referents of pointers are not permitted to contain references.

Some examples of pointer types are given in the table below:

Example Description
byte* Pointer to byte
char* Pointer to char
int** Pointer to pointer to int
int*[] Single-dimensional array of pointers to int
void* Pointer to unknown type

For a given implementation, all pointer types must have the same size and representation.

Unlike C and C++, when multiple pointers are declared in the same declaration, in C# the * is written along with the underlying type only, not as a prefix punctuator on each pointer name. For example

int* pi, pj;    // NOT as int *pi, *pj;

The value of a pointer having type T* represents the address of a variable of type T. The pointer indirection operator * (Pointer indirection) may be used to access this variable. For example, given a variable P of type int*, the expression *P denotes the int variable found at the address contained in P.

Like an object reference, a pointer may be null. Applying the indirection operator to a null pointer results in implementation-defined behavior. A pointer with value null is represented by all-bits-zero.

The void* type represents a pointer to an unknown type. Because the referent type is unknown, the indirection operator cannot be applied to a pointer of type void*, nor can any arithmetic be performed on such a pointer. However, a pointer of type void* can be cast to any other pointer type (and vice versa).

Pointer types are a separate category of types. Unlike reference types and value types, pointer types do not inherit from object and no conversions exist between pointer types and object. In particular, boxing and unboxing (Boxing and unboxing) are not supported for pointers. However, conversions are permitted between different pointer types and between pointer types and the integral types. This is described in Pointer conversions.

A pointer_type cannot be used as a type argument (Constructed types), and type inference (Type inference) fails on generic method calls that would have inferred a type argument to be a pointer type.

A pointer_type may be used as the type of a volatile field (Volatile fields).

Although pointers can be passed as ref or out parameters, doing so can cause undefined behavior, since the pointer may well be set to point to a local variable which no longer exists when the called method returns, or the fixed object to which it used to point, is no longer fixed. For example:

using System;

class Test
{
    static int value = 20;

    unsafe static void F(out int* pi1, ref int* pi2) {
        int i = 10;
        pi1 = &i;

        fixed (int* pj = &value) {
            // ...
            pi2 = pj;
        }
    }

    static void Main() {
        int i = 10;
        unsafe {
            int* px1;
            int* px2 = &i;

            F(out px1, ref px2);

            Console.WriteLine("*px1 = {0}, *px2 = {1}",
                *px1, *px2);    // undefined behavior
        }
    }
}

A method can return a value of some type, and that type can be a pointer. For example, when given a pointer to a contiguous sequence of ints, that sequence's element count, and some other int value, the following method returns the address of that value in that sequence, if a match occurs; otherwise it returns null:

unsafe static int* Find(int* pi, int size, int value) {
    for (int i = 0; i < size; ++i) {
        if (*pi == value) 
            return pi;
        ++pi;
    }
    return null;
}

In an unsafe context, several constructs are available for operating on pointers:

Fixed and moveable variables

The address-of operator (The address-of operator) and the fixed statement (The fixed statement) divide variables into two categories: Fixed variables and moveable variables.

Fixed variables reside in storage locations that are unaffected by operation of the garbage collector. (Examples of fixed variables include local variables, value parameters, and variables created by dereferencing pointers.) On the other hand, moveable variables reside in storage locations that are subject to relocation or disposal by the garbage collector. (Examples of moveable variables include fields in objects and elements of arrays.)

The & operator (The address-of operator) permits the address of a fixed variable to be obtained without restrictions. However, because a moveable variable is subject to relocation or disposal by the garbage collector, the address of a moveable variable can only be obtained using a fixed statement (The fixed statement), and that address remains valid only for the duration of that fixed statement.

In precise terms, a fixed variable is one of the following:

  • A variable resulting from a simple_name (Simple names) that refers to a local variable or a value parameter, unless the variable is captured by an anonymous function.
  • A variable resulting from a member_access (Member access) of the form V.I, where V is a fixed variable of a struct_type.
  • A variable resulting from a pointer_indirection_expression (Pointer indirection) of the form *P, a pointer_member_access (Pointer member access) of the form P->I, or a pointer_element_access (Pointer element access) of the form P[E].

All other variables are classified as moveable variables.

Note that a static field is classified as a moveable variable. Also note that a ref or out parameter is classified as a moveable variable, even if the argument given for the parameter is a fixed variable. Finally, note that a variable produced by dereferencing a pointer is always classified as a fixed variable.

Pointer conversions

In an unsafe context, the set of available implicit conversions (Implicit conversions) is extended to include the following implicit pointer conversions:

  • From any pointer_type to the type void*.
  • From the null literal to any pointer_type.

Additionally, in an unsafe context, the set of available explicit conversions (Explicit conversions) is extended to include the following explicit pointer conversions:

  • From any pointer_type to any other pointer_type.
  • From sbyte, byte, short, ushort, int, uint, long, or ulong to any pointer_type.
  • From any pointer_type to sbyte, byte, short, ushort, int, uint, long, or ulong.

Finally, in an unsafe context, the set of standard implicit conversions (Standard implicit conversions) includes the following pointer conversion:

  • From any pointer_type to the type void*.

Conversions between two pointer types never change the actual pointer value. In other words, a conversion from one pointer type to another has no effect on the underlying address given by the pointer.

When one pointer type is converted to another, if the resulting pointer is not correctly aligned for the pointed-to type, the behavior is undefined if the result is dereferenced. In general, the concept "correctly aligned" is transitive: if a pointer to type A is correctly aligned for a pointer to type B, which, in turn, is correctly aligned for a pointer to type C, then a pointer to type A is correctly aligned for a pointer to type C.

Consider the following case in which a variable having one type is accessed via a pointer to a different type:

char c = 'A';
char* pc = &c;
void* pv = pc;
int* pi = (int*)pv;
int i = *pi;         // undefined
*pi = 123456;        // undefined

When a pointer type is converted to a pointer to byte, the result points to the lowest addressed byte of the variable. Successive increments of the result, up to the size of the variable, yield pointers to the remaining bytes of that variable. For example, the following method displays each of the eight bytes in a double as a hexadecimal value:

using System;

class Test
{
    unsafe static void Main() {
      double d = 123.456e23;
        unsafe {
           byte* pb = (byte*)&d;
            for (int i = 0; i < sizeof(double); ++i)
               Console.Write("{0:X2} ", *pb++);
            Console.WriteLine();
        }
    }
}

Of course, the output produced depends on endianness.

Mappings between pointers and integers are implementation-defined. However, on 32* and 64-bit CPU architectures with a linear address space, conversions of pointers to or from integral types typically behave exactly like conversions of uint or ulong values, respectively, to or from those integral types.

Pointer arrays

In an unsafe context, arrays of pointers can be constructed. Only some of the conversions that apply to other array types are allowed on pointer arrays:

  • The implicit reference conversion (Implicit reference conversions) from any array_type to System.Array and the interfaces it implements also applies to pointer arrays. However, any attempt to access the array elements through System.Array or the interfaces it implements will result in an exception at run-time, as pointer types are not convertible to object.
  • The implicit and explicit reference conversions (Implicit reference conversions, Explicit reference conversions) from a single-dimensional array type S[] to System.Collections.Generic.IList<T> and its generic base interfaces never apply to pointer arrays, since pointer types cannot be used as type arguments, and there are no conversions from pointer types to non-pointer types.
  • The explicit reference conversion (Explicit reference conversions) from System.Array and the interfaces it implements to any array_type applies to pointer arrays.
  • The explicit reference conversions (Explicit reference conversions) from System.Collections.Generic.IList<S> and its base interfaces to a single-dimensional array type T[] never applies to pointer arrays, since pointer types cannot be used as type arguments, and there are no conversions from pointer types to non-pointer types.

These restrictions mean that the expansion for the foreach statement over arrays described in The foreach statement cannot be applied to pointer arrays. Instead, a foreach statement of the form

foreach (V v in x) embedded_statement

where the type of x is an array type of the form T[,,...,], N is the number of dimensions minus 1 and T or V is a pointer type, is expanded using nested for-loops as follows:

{
    T[,,...,] a = x;
    for (int i0 = a.GetLowerBound(0); i0 <= a.GetUpperBound(0); i0++)
    for (int i1 = a.GetLowerBound(1); i1 <= a.GetUpperBound(1); i1++)
    ...
    for (int iN = a.GetLowerBound(N); iN <= a.GetUpperBound(N); iN++) {
        V v = (V)a.GetValue(i0,i1,...,iN);
        embedded_statement
    }
}

The variables a, i0, i1, ..., iN are not visible to or accessible to x or the embedded_statement or any other source code of the program. The variable v is read-only in the embedded statement. If there is not an explicit conversion (Pointer conversions) from T (the element type) to V, an error is produced and no further steps are taken. If x has the value null, a System.NullReferenceException is thrown at run-time.

Pointers in expressions

In an unsafe context, an expression may yield a result of a pointer type, but outside an unsafe context it is a compile-time error for an expression to be of a pointer type. In precise terms, outside an unsafe context a compile-time error occurs if any simple_name (Simple names), member_access (Member access), invocation_expression (Invocation expressions), or element_access (Element access) is of a pointer type.

In an unsafe context, the primary_no_array_creation_expression (Primary expressions) and unary_expression (Unary operators) productions permit the following additional constructs:

primary_no_array_creation_expression_unsafe
    : pointer_member_access
    | pointer_element_access
    | sizeof_expression
    ;

unary_expression_unsafe
    : pointer_indirection_expression
    | addressof_expression
    ;

These constructs are described in the following sections. The precedence and associativity of the unsafe operators is implied by the grammar.

Pointer indirection

A pointer_indirection_expression consists of an asterisk (*) followed by a unary_expression.

pointer_indirection_expression
    : '*' unary_expression
    ;

The unary * operator denotes pointer indirection and is used to obtain the variable to which a pointer points. The result of evaluating *P, where P is an expression of a pointer type T*, is a variable of type T. It is a compile-time error to apply the unary * operator to an expression of type void* or to an expression that isn't of a pointer type.

The effect of applying the unary * operator to a null pointer is implementation-defined. In particular, there is no guarantee that this operation throws a System.NullReferenceException.

If an invalid value has been assigned to the pointer, the behavior of the unary * operator is undefined. Among the invalid values for dereferencing a pointer by the unary * operator are an address inappropriately aligned for the type pointed to (see example in Pointer conversions), and the address of a variable after the end of its lifetime.

For purposes of definite assignment analysis, a variable produced by evaluating an expression of the form *P is considered initially assigned (Initially assigned variables).

Pointer member access

A pointer_member_access consists of a primary_expression, followed by a "->" token, followed by an identifier and an optional type_argument_list.

pointer_member_access
    : primary_expression '->' identifier
    ;

In a pointer member access of the form P->I, P must be an expression of a pointer type other than void*, and I must denote an accessible member of the type to which P points.

A pointer member access of the form P->I is evaluated exactly as (*P).I. For a description of the pointer indirection operator (*), see Pointer indirection. For a description of the member access operator (.), see Member access.

In the example

using System;

struct Point
{
    public int x;
    public int y;

    public override string ToString() {
        return "(" + x + "," + y + ")";
    }
}

class Test
{
    static void Main() {
        Point point;
        unsafe {
            Point* p = &point;
            p->x = 10;
            p->y = 20;
            Console.WriteLine(p->ToString());
        }
    }
}

the -> operator is used to access fields and invoke a method of a struct through a pointer. Because the operation P->I is precisely equivalent to (*P).I, the Main method could equally well have been written:

class Test
{
    static void Main() {
        Point point;
        unsafe {
            Point* p = &point;
            (*p).x = 10;
            (*p).y = 20;
            Console.WriteLine((*p).ToString());
        }
    }
}

Pointer element access

A pointer_element_access consists of a primary_no_array_creation_expression followed by an expression enclosed in "[" and "]".

pointer_element_access
    : primary_no_array_creation_expression '[' expression ']'
    ;

In a pointer element access of the form P[E], P must be an expression of a pointer type other than void*, and E must be an expression that can be implicitly converted to int, uint, long, or ulong.

A pointer element access of the form P[E] is evaluated exactly as *(P + E). For a description of the pointer indirection operator (*), see Pointer indirection. For a description of the pointer addition operator (+), see Pointer arithmetic.

In the example

class Test
{
    static void Main() {
        unsafe {
            char* p = stackalloc char[256];
            for (int i = 0; i < 256; i++) p[i] = (char)i;
        }
    }
}

a pointer element access is used to initialize the character buffer in a for loop. Because the operation P[E] is precisely equivalent to *(P + E), the example could equally well have been written:

class Test
{
    static void Main() {
        unsafe {
            char* p = stackalloc char[256];
            for (int i = 0; i < 256; i++) *(p + i) = (char)i;
        }
    }
}

The pointer element access operator does not check for out-of-bounds errors and the behavior when accessing an out-of-bounds element is undefined. This is the same as C and C++.

The address-of operator

An addressof_expression consists of an ampersand (&) followed by a unary_expression.

addressof_expression
    : '&' unary_expression
    ;

Given an expression E which is of a type T and is classified as a fixed variable (Fixed and moveable variables), the construct &E computes the address of the variable given by E. The type of the result is T* and is classified as a value. A compile-time error occurs if E is not classified as a variable, if E is classified as a read-only local variable, or if E denotes a moveable variable. In the last case, a fixed statement (The fixed statement) can be used to temporarily "fix" the variable before obtaining its address. As stated in Member access, outside an instance constructor or static constructor for a struct or class that defines a readonly field, that field is considered a value, not a variable. As such, its address cannot be taken. Similarly, the address of a constant cannot be taken.

The & operator does not require its argument to be definitely assigned, but following an & operation, the variable to which the operator is applied is considered definitely assigned in the execution path in which the operation occurs. It is the responsibility of the programmer to ensure that correct initialization of the variable actually does take place in this situation.

In the example

using System;

class Test
{
    static void Main() {
        int i;
        unsafe {
            int* p = &i;
            *p = 123;
        }
        Console.WriteLine(i);
    }
}

i is considered definitely assigned following the &i operation used to initialize p. The assignment to *p in effect initializes i, but the inclusion of this initialization is the responsibility of the programmer, and no compile-time error would occur if the assignment was removed.

The rules of definite assignment for the & operator exist such that redundant initialization of local variables can be avoided. For example, many external APIs take a pointer to a structure which is filled in by the API. Calls to such APIs typically pass the address of a local struct variable, and without the rule, redundant initialization of the struct variable would be required.

Pointer increment and decrement

In an unsafe context, the ++ and -- operators (Postfix increment and decrement operators and Prefix increment and decrement operators) can be applied to pointer variables of all types except void*. Thus, for every pointer type T*, the following operators are implicitly defined:

T* operator ++(T* x);
T* operator --(T* x);

The operators produce the same results as x + 1 and x - 1, respectively (Pointer arithmetic). In other words, for a pointer variable of type T*, the ++ operator adds sizeof(T) to the address contained in the variable, and the -- operator subtracts sizeof(T) from the address contained in the variable.

If a pointer increment or decrement operation overflows the domain of the pointer type, the result is implementation-defined, but no exceptions are produced.

Pointer arithmetic

In an unsafe context, the + and - operators (Addition operator and Subtraction operator) can be applied to values of all pointer types except void*. Thus, for every pointer type T*, the following operators are implicitly defined:

T* operator +(T* x, int y);
T* operator +(T* x, uint y);
T* operator +(T* x, long y);
T* operator +(T* x, ulong y);

T* operator +(int x, T* y);
T* operator +(uint x, T* y);
T* operator +(long x, T* y);
T* operator +(ulong x, T* y);

T* operator -(T* x, int y);
T* operator -(T* x, uint y);
T* operator -(T* x, long y);
T* operator -(T* x, ulong y);

long operator -(T* x, T* y);

Given an expression P of a pointer type T* and an expression N of type int, uint, long, or ulong, the expressions P + N and N + P compute the pointer value of type T* that results from adding N * sizeof(T) to the address given by P. Likewise, the expression P - N computes the pointer value of type T* that results from subtracting N * sizeof(T) from the address given by P.

Given two expressions, P and Q, of a pointer type T*, the expression P - Q computes the difference between the addresses given by P and Q and then divides that difference by sizeof(T). The type of the result is always long. In effect, P - Q is computed as ((long)(P) - (long)(Q)) / sizeof(T).

For example:

using System;

class Test
{
    static void Main() {
        unsafe {
            int* values = stackalloc int[20];
            int* p = &values[1];
            int* q = &values[15];
            Console.WriteLine("p - q = {0}", p - q);
            Console.WriteLine("q - p = {0}", q - p);
        }
    }
}

which produces the output:

p - q = -14
q - p = 14

If a pointer arithmetic operation overflows the domain of the pointer type, the result is truncated in an implementation-defined fashion, but no exceptions are produced.

Pointer comparison

In an unsafe context, the ==, !=, <, >, <=, and => operators (Relational and type-testing operators) can be applied to values of all pointer types. The pointer comparison operators are:

bool operator ==(void* x, void* y);
bool operator !=(void* x, void* y);
bool operator <(void* x, void* y);
bool operator >(void* x, void* y);
bool operator <=(void* x, void* y);
bool operator >=(void* x, void* y);

Because an implicit conversion exists from any pointer type to the void* type, operands of any pointer type can be compared using these operators. The comparison operators compare the addresses given by the two operands as if they were unsigned integers.

The sizeof operator

The sizeof operator returns the number of bytes occupied by a variable of a given type. The type specified as an operand to sizeof must be an unmanaged_type (Pointer types).

sizeof_expression
    : 'sizeof' '(' unmanaged_type ')'
    ;

The result of the sizeof operator is a value of type int. For certain predefined types, the sizeof operator yields a constant value as shown in the table below.

Expression Result
sizeof(sbyte) 1
sizeof(byte) 1
sizeof(short) 2
sizeof(ushort) 2
sizeof(int) 4
sizeof(uint) 4
sizeof(long) 8
sizeof(ulong) 8
sizeof(char) 2
sizeof(float) 4
sizeof(double) 8
sizeof(bool) 1

For all other types, the result of the sizeof operator is implementation-defined and is classified as a value, not a constant.

The order in which members are packed into a struct is unspecified.

For alignment purposes, there may be unnamed padding at the beginning of a struct, within a struct, and at the end of the struct. The contents of the bits used as padding are indeterminate.

When applied to an operand that has struct type, the result is the total number of bytes in a variable of that type, including any padding.

The fixed statement

In an unsafe context, the embedded_statement (Statements) production permits an additional construct, the fixed statement, which is used to "fix" a moveable variable such that its address remains constant for the duration of the statement.

fixed_statement
    : 'fixed' '(' pointer_type fixed_pointer_declarators ')' embedded_statement
    ;

fixed_pointer_declarators
    : fixed_pointer_declarator (','  fixed_pointer_declarator)*
    ;

fixed_pointer_declarator
    : identifier '=' fixed_pointer_initializer
    ;

fixed_pointer_initializer
    : '&' variable_reference
    | expression
    ;

Each fixed_pointer_declarator declares a local variable of the given pointer_type and initializes that local variable with the address computed by the corresponding fixed_pointer_initializer. A local variable declared in a fixed statement is accessible in any fixed_pointer_initializers occurring to the right of that variable's declaration, and in the embedded_statement of the fixed statement. A local variable declared by a fixed statement is considered read-only. A compile-time error occurs if the embedded statement attempts to modify this local variable (via assignment or the ++ and -- operators) or pass it as a ref or out parameter.

A fixed_pointer_initializer can be one of the following:

  • The token "&" followed by a variable_reference (Precise rules for determining definite assignment) to a moveable variable (Fixed and moveable variables) of an unmanaged type T, provided the type T* is implicitly convertible to the pointer type given in the fixed statement. In this case, the initializer computes the address of the given variable, and the variable is guaranteed to remain at a fixed address for the duration of the fixed statement.
  • An expression of an array_type with elements of an unmanaged type T, provided the type T* is implicitly convertible to the pointer type given in the fixed statement. In this case, the initializer computes the address of the first element in the array, and the entire array is guaranteed to remain at a fixed address for the duration of the fixed statement. If the array expression is null or if the array has zero elements, the initializer computes an address equal to zero.
  • An expression of type string, provided the type char* is implicitly convertible to the pointer type given in the fixed statement. In this case, the initializer computes the address of the first character in the string, and the entire string is guaranteed to remain at a fixed address for the duration of the fixed statement. The behavior of the fixed statement is implementation-defined if the string expression is null.
  • A simple_name or member_access that references a fixed size buffer member of a moveable variable, provided the type of the fixed size buffer member is implicitly convertible to the pointer type given in the fixed statement. In this case, the initializer computes a pointer to the first element of the fixed size buffer (Fixed size buffers in expressions), and the fixed size buffer is guaranteed to remain at a fixed address for the duration of the fixed statement.

For each address computed by a fixed_pointer_initializer the fixed statement ensures that the variable referenced by the address is not subject to relocation or disposal by the garbage collector for the duration of the fixed statement. For example, if the address computed by a fixed_pointer_initializer references a field of an object or an element of an array instance, the fixed statement guarantees that the containing object instance is not relocated or disposed of during the lifetime of the statement.

It is the programmer's responsibility to ensure that pointers created by fixed statements do not survive beyond execution of those statements. For example, when pointers created by fixed statements are passed to external APIs, it is the programmer's responsibility to ensure that the APIs retain no memory of these pointers.

Fixed objects may cause fragmentation of the heap (because they can't be moved). For that reason, objects should be fixed only when absolutely necessary and then only for the shortest amount of time possible.

The example

class Test
{
    static int x;
    int y;

    unsafe static void F(int* p) {
        *p = 1;
    }

    static void Main() {
        Test t = new Test();
        int[] a = new int[10];
        unsafe {
            fixed (int* p = &x) F(p);
            fixed (int* p = &t.y) F(p);
            fixed (int* p = &a[0]) F(p);
            fixed (int* p = a) F(p);
        }
    }
}

demonstrates several uses of the fixed statement. The first statement fixes and obtains the address of a static field, the second statement fixes and obtains the address of an instance field, and the third statement fixes and obtains the address of an array element. In each case it would have been an error to use the regular & operator since the variables are all classified as moveable variables.

The fourth fixed statement in the example above produces a similar result to the third.

This example of the fixed statement uses string:

class Test
{
    static string name = "xx";

    unsafe static void F(char* p) {
        for (int i = 0; p[i] != '\0'; ++i)
            Console.WriteLine(p[i]);
    }

    static void Main() {
        unsafe {
            fixed (char* p = name) F(p);
            fixed (char* p = "xx") F(p);
        }
    }
}

In an unsafe context array elements of single-dimensional arrays are stored in increasing index order, starting with index 0 and ending with index Length - 1. For multi-dimensional arrays, array elements are stored such that the indices of the rightmost dimension are increased first, then the next left dimension, and so on to the left. Within a fixed statement that obtains a pointer p to an array instance a, the pointer values ranging from p to p + a.Length - 1 represent addresses of the elements in the array. Likewise, the variables ranging from p[0] to p[a.Length - 1] represent the actual array elements. Given the way in which arrays are stored, we can treat an array of any dimension as though it were linear.

For example:

using System;

class Test
{
    static void Main() {
        int[,,] a = new int[2,3,4];
        unsafe {
            fixed (int* p = a) {
                for (int i = 0; i < a.Length; ++i)    // treat as linear
                    p[i] = i;
            }
        }

        for (int i = 0; i < 2; ++i)
            for (int j = 0; j < 3; ++j) {
                for (int k = 0; k < 4; ++k)
                    Console.Write("[{0},{1},{2}] = {3,2} ", i, j, k, a[i,j,k]);
                Console.WriteLine();
            }
    }
}

which produces the output:

[0,0,0] =  0 [0,0,1] =  1 [0,0,2] =  2 [0,0,3] =  3
[0,1,0] =  4 [0,1,1] =  5 [0,1,2] =  6 [0,1,3] =  7
[0,2,0] =  8 [0,2,1] =  9 [0,2,2] = 10 [0,2,3] = 11
[1,0,0] = 12 [1,0,1] = 13 [1,0,2] = 14 [1,0,3] = 15
[1,1,0] = 16 [1,1,1] = 17 [1,1,2] = 18 [1,1,3] = 19
[1,2,0] = 20 [1,2,1] = 21 [1,2,2] = 22 [1,2,3] = 23

In the example

class Test
{
    unsafe static void Fill(int* p, int count, int value) {
        for (; count != 0; count--) *p++ = value;
    }

    static void Main() {
        int[] a = new int[100];
        unsafe {
            fixed (int* p = a) Fill(p, 100, -1);
        }
    }
}

a fixed statement is used to fix an array so its address can be passed to a method that takes a pointer.

In the example:

unsafe struct Font
{
    public int size;
    public fixed char name[32];
}

class Test
{
    unsafe static void PutString(string s, char* buffer, int bufSize) {
        int len = s.Length;
        if (len > bufSize) len = bufSize;
        for (int i = 0; i < len; i++) buffer[i] = s[i];
        for (int i = len; i < bufSize; i++) buffer[i] = (char)0;
    }

    Font f;

    unsafe static void Main()
    {
        Test test = new Test();
        test.f.size = 10;
        fixed (char* p = test.f.name) {
            PutString("Times New Roman", p, 32);
        }
    }
}

a fixed statement is used to fix a fixed size buffer of a struct so its address can be used as a pointer.

A char* value produced by fixing a string instance always points to a null-terminated string. Within a fixed statement that obtains a pointer p to a string instance s, the pointer values ranging from p to p + s.Length - 1 represent addresses of the characters in the string, and the pointer value p + s.Length always points to a null character (the character with value '\0').

Modifying objects of managed type through fixed pointers can results in undefined behavior. For example, because strings are immutable, it is the programmer's responsibility to ensure that the characters referenced by a pointer to a fixed string are not modified.

The automatic null-termination of strings is particularly convenient when calling external APIs that expect "C-style" strings. Note, however, that a string instance is permitted to contain null characters. If such null characters are present, the string will appear truncated when treated as a null-terminated char*.

Fixed size buffers

Fixed size buffers are used to declare "C style" in-line arrays as members of structs, and are primarily useful for interfacing with unmanaged APIs.

Fixed size buffer declarations

A fixed size buffer is a member that represents storage for a fixed length buffer of variables of a given type. A fixed size buffer declaration introduces one or more fixed size buffers of a given element type. Fixed size buffers are only permitted in struct declarations and can only occur in unsafe contexts (Unsafe contexts).

struct_member_declaration_unsafe
    : fixed_size_buffer_declaration
    ;

fixed_size_buffer_declaration
    : attributes? fixed_size_buffer_modifier* 'fixed' buffer_element_type fixed_size_buffer_declarator+ ';'
    ;

fixed_size_buffer_modifier
    : 'new'
    | 'public'
    | 'protected'
    | 'internal'
    | 'private'
    | 'unsafe'
    ;

buffer_element_type
    : type
    ;

fixed_size_buffer_declarator
    : identifier '[' constant_expression ']'
    ;

A fixed size buffer declaration may include a set of attributes (Attributes), a new modifier (Modifiers), a valid combination of the four access modifiers (Type parameters and constraints) and an unsafe modifier (Unsafe contexts). The attributes and modifiers apply to all of the members declared by the fixed size buffer declaration. It is an error for the same modifier to appear multiple times in a fixed size buffer declaration.

A fixed size buffer declaration is not permitted to include the static modifier.

The buffer element type of a fixed size buffer declaration specifies the element type of the buffer(s) introduced by the declaration. The buffer element type must be one of the predefined types sbyte, byte, short, ushort, int, uint, long, ulong, char, float, double, or bool.

The buffer element type is followed by a list of fixed size buffer declarators, each of which introduces a new member. A fixed size buffer declarator consists of an identifier that names the member, followed by a constant expression enclosed in [ and ] tokens. The constant expression denotes the number of elements in the member introduced by that fixed size buffer declarator. The type of the constant expression must be implicitly convertible to type int, and the value must be a non-zero positive integer.

The elements of a fixed size buffer are guaranteed to be laid out sequentially in memory.

A fixed size buffer declaration that declares multiple fixed size buffers is equivalent to multiple declarations of a single fixed size buffer declaration with the same attributes, and element types. For example

unsafe struct A
{
   public fixed int x[5], y[10], z[100];
}

is equivalent to

unsafe struct A
{
   public fixed int x[5];
   public fixed int y[10];
   public fixed int z[100];
}

Fixed size buffers in expressions

Member lookup (Operators) of a fixed size buffer member proceeds exactly like member lookup of a field.

A fixed size buffer can be referenced in an expression using a simple_name (Type inference) or a member_access (Compile-time checking of dynamic overload resolution).

When a fixed size buffer member is referenced as a simple name, the effect is the same as a member access of the form this.I, where I is the fixed size buffer member.

In a member access of the form E.I, if E is of a struct type and a member lookup of I in that struct type identifies a fixed size member, then E.I is evaluated an classified as follows:

  • If the expression E.I does not occur in an unsafe context, a compile-time error occurs.
  • If E is classified as a value, a compile-time error occurs.
  • Otherwise, if E is a moveable variable (Fixed and moveable variables) and the expression E.I is not a fixed_pointer_initializer (The fixed statement), a compile-time error occurs.
  • Otherwise, E references a fixed variable and the result of the expression is a pointer to the first element of the fixed size buffer member I in E. The result is of type S*, where S is the element type of I, and is classified as a value.

The subsequent elements of the fixed size buffer can be accessed using pointer operations from the first element. Unlike access to arrays, access to the elements of a fixed size buffer is an unsafe operation and is not range checked.

The following example declares and uses a struct with a fixed size buffer member.

unsafe struct Font
{
    public int size;
    public fixed char name[32];
}

class Test
{
    unsafe static void PutString(string s, char* buffer, int bufSize) {
        int len = s.Length;
        if (len > bufSize) len = bufSize;
        for (int i = 0; i < len; i++) buffer[i] = s[i];
        for (int i = len; i < bufSize; i++) buffer[i] = (char)0;
    }

    unsafe static void Main()
    {
        Font f;
        f.size = 10;
        PutString("Times New Roman", f.name, 32);
    }
}

Definite assignment checking

Fixed size buffers are not subject to definite assignment checking (Definite assignment), and fixed size buffer members are ignored for purposes of definite assignment checking of struct type variables.

When the outermost containing struct variable of a fixed size buffer member is a static variable, an instance variable of a class instance, or an array element, the elements of the fixed size buffer are automatically initialized to their default values (Default values). In all other cases, the initial content of a fixed size buffer is undefined.

Stack allocation

In an unsafe context, a local variable declaration (Local variable declarations) may include a stack allocation initializer which allocates memory from the call stack.

local_variable_initializer_unsafe
    : stackalloc_initializer
    ;

stackalloc_initializer
    : 'stackalloc' unmanaged_type '[' expression ']'
    ;

The unmanaged_type indicates the type of the items that will be stored in the newly allocated location, and the expression indicates the number of these items. Taken together, these specify the required allocation size. Since the size of a stack allocation cannot be negative, it is a compile-time error to specify the number of items as a constant_expression that evaluates to a negative value.

A stack allocation initializer of the form stackalloc T[E] requires T to be an unmanaged type (Pointer types) and E to be an expression of type int. The construct allocates E * sizeof(T) bytes from the call stack and returns a pointer, of type T*, to the newly allocated block. If E is a negative value, then the behavior is undefined. If E is zero, then no allocation is made, and the pointer returned is implementation-defined. If there is not enough memory available to allocate a block of the given size, a System.StackOverflowException is thrown.

The content of the newly allocated memory is undefined.

Stack allocation initializers are not permitted in catch or finally blocks (The try statement).

There is no way to explicitly free memory allocated using stackalloc. All stack allocated memory blocks created during the execution of a function member are automatically discarded when that function member returns. This corresponds to the alloca function, an extension commonly found in C and C++ implementations.

In the example

using System;

class Test
{
    static string IntToString(int value) {
        int n = value >= 0? value: -value;
        unsafe {
            char* buffer = stackalloc char[16];
            char* p = buffer + 16;
            do {
                *--p = (char)(n % 10 + '0');
                n /= 10;
            } while (n != 0);
            if (value < 0) *--p = '-';
            return new string(p, 0, (int)(buffer + 16 - p));
        }
    }

    static void Main() {
        Console.WriteLine(IntToString(12345));
        Console.WriteLine(IntToString(-999));
    }
}

a stackalloc initializer is used in the IntToString method to allocate a buffer of 16 characters on the stack. The buffer is automatically discarded when the method returns.

Dynamic memory allocation

Except for the stackalloc operator, C# provides no predefined constructs for managing non-garbage collected memory. Such services are typically provided by supporting class libraries or imported directly from the underlying operating system. For example, the Memory class below illustrates how the heap functions of an underlying operating system might be accessed from C#:

using System;
using System.Runtime.InteropServices;

public unsafe class Memory
{
    // Handle for the process heap. This handle is used in all calls to the
    // HeapXXX APIs in the methods below.
    static int ph = GetProcessHeap();

    // Private instance constructor to prevent instantiation.
    private Memory() {}

    // Allocates a memory block of the given size. The allocated memory is
    // automatically initialized to zero.
    public static void* Alloc(int size) {
        void* result = HeapAlloc(ph, HEAP_ZERO_MEMORY, size);
        if (result == null) throw new OutOfMemoryException();
        return result;
    }

    // Copies count bytes from src to dst. The source and destination
    // blocks are permitted to overlap.
    public static void Copy(void* src, void* dst, int count) {
        byte* ps = (byte*)src;
        byte* pd = (byte*)dst;
        if (ps > pd) {
            for (; count != 0; count--) *pd++ = *ps++;
        }
        else if (ps < pd) {
            for (ps += count, pd += count; count != 0; count--) *--pd = *--ps;
        }
    }

    // Frees a memory block.
    public static void Free(void* block) {
        if (!HeapFree(ph, 0, block)) throw new InvalidOperationException();
    }

    // Re-allocates a memory block. If the reallocation request is for a
    // larger size, the additional region of memory is automatically
    // initialized to zero.
    public static void* ReAlloc(void* block, int size) {
        void* result = HeapReAlloc(ph, HEAP_ZERO_MEMORY, block, size);
        if (result == null) throw new OutOfMemoryException();
        return result;
    }

    // Returns the size of a memory block.
    public static int SizeOf(void* block) {
        int result = HeapSize(ph, 0, block);
        if (result == -1) throw new InvalidOperationException();
        return result;
    }

    // Heap API flags
    const int HEAP_ZERO_MEMORY = 0x00000008;

    // Heap API functions
    [DllImport("kernel32")]
    static extern int GetProcessHeap();

    [DllImport("kernel32")]
    static extern void* HeapAlloc(int hHeap, int flags, int size);

    [DllImport("kernel32")]
    static extern bool HeapFree(int hHeap, int flags, void* block);

    [DllImport("kernel32")]
    static extern void* HeapReAlloc(int hHeap, int flags, void* block, int size);

    [DllImport("kernel32")]
    static extern int HeapSize(int hHeap, int flags, void* block);
}

An example that uses the Memory class is given below:

class Test
{
    static void Main() {
        unsafe {
            byte* buffer = (byte*)Memory.Alloc(256);
            try {
                for (int i = 0; i < 256; i++) buffer[i] = (byte)i;
                byte[] array = new byte[256];
                fixed (byte* p = array) Memory.Copy(buffer, p, 256); 
            }
            finally {
                Memory.Free(buffer);
            }
            for (int i = 0; i < 256; i++) Console.WriteLine(array[i]);
        }
    }
}

The example allocates 256 bytes of memory through Memory.Alloc and initializes the memory block with values increasing from 0 to 255. It then allocates a 256 element byte array and uses Memory.Copy to copy the contents of the memory block into the byte array. Finally, the memory block is freed using Memory.Free and the contents of the byte array are output on the console.