# Following Specification

At this point of time I would like to look back and refer to C11 specification n1570.pdf, which, is the C11 specification with me. Threading is not yet implemented in any compiler so we will have to wait for compilers to support it before we can study it. I will try to follow the organization of content given in the specification. This treatment of C99 is very tightly coupled with compiler(gcc) and specification. If someone says that a book can be written by ignoring any of these two then I will be the last person to believe that.:P However, note that I will not include all the contents of specification and will exclude some of very obvious and trivial sections/subsections. The following terms, definitions and symbols have come from specification, however, some are omitted for the sake of conciseness. When I will repeat the specification at times I will do a verbatim copy just for quick reference and then add an explanation to that. Believe me you can learn most of the language just by studying the specification itself. Chapter no. 2, 3 and 4 will have material which may have redundant material discussed elsewhere in the book. Note that specification is directed more at compiler writers rather than at developers. However, it still covers a lot of ground and is also helpful to understand the nature of different compilers as where they can go in different directions.

## Terms, Definitions and Symbols

1. access: <execution-time action> to read or modify the value of an object.

NOTE 1: Where only one of these two actions is meant, “read” or “modify” is used.

NOTE 2: “Modify” includes the case where the new value being stored is the same as the previous value.

NOTE 3: Expressions that are not evaluated do not access objects.

There are two parts of any program. Data and instruction. Programs are stored in file on some non-volatile storage for example, hard disk drive, CD, DVD, tape drive. When they are executed from non-volatile storage they are transferred to some volatile storage typically RAM (Random Access Memory) of the computer. When a program is executed it becomes a living entity capable of doing something and sometimes also referred as process. So when the contents of RAM (henceforth referred as memory) is either read or written (it does not matter whether the value is same or new) to then it is defined as access. Here point to be noted is that the expressions which will not be evaluated do not access objects.

There are certain issues wit the term read and modify. One can ask it it the case that multiplication or division of a number with 1 modifies it or addition or subtraction of a number with 0 modifies it. The answer is yes on all these accounts. Similarly, you can ask if a bit-field is accessed then is it the case that the bit-fields sharing the storage in memory are also accessed. My answer would be yes. The reason is there is no way to get address of an individual bit and we get at least 32-bits in one fetch cycle on a 32-bit processor.

There can be various reasons why an expression is not evaluated. For example:

• being part of a statement which is not executed probably because of conditional compilation,

• being part of a sub-expression whose evaluation is conditional on other sub-expressions within a full expression; this means the above point, and

• being an operand of the sizeof operator. sizeof operator never evaluates its operand or try to access them. It just computes their sizes and pass it along.

Also, consider the following program:

#include <stdio.h>

int main()
{
int a=0, b=0, c=3, d=4;

a = c + d;
b = c + d;

return 0;
}


In this code it is not necessary that data for c and d will be accessed for second addition.

Also consider the following program:

#include <stdio.h>

typedef struct {
volatile int x1:3;
volatile int x2:3;
volatile int x3:4;
volatile int x4:1;
} S;

int main()
{
int i = 7;
S s;

i = s.x2;

return 0;
}


Now depending on whether system is little-endian or big-endian the storage of x2 and x3 will be in different bytes. Also, this will change the way bytes are accessed when x2 is referred in memory. Many combinations are possible depending on processor architecture.

1. alignment: requirement that objects of a particular type be located on storage boundaries with addresses that are particular multiples of a byte address.

Say your program requires x bytes of memory then it will not be always given x bytes but something more. Say an object requires y bytes then it will be always greater or equal to y bytes. This is required so that objects are always located on storage boundaries that are particular multiples of byte address. The reason for this alignment lies in the efficiency of the operating system as a whole. As we know that on 32-bit systems data bus is 32 bits similarly on 64-bit systems it is 64 bits. This means in one fetch cycle (read up on this on some microprocessor or computer architecture book) only 32-bits can be fetched. 32-bits means 4 bytes. Oops! I did not tell you about bits and bytes conversion. Not even nibble. However, since 4 bytes can be fetched in one cycle compiler tries to optimize the data in group of 4 bytes. Given below are some examples.

/* Description: Demonstration of structure padding and memory alignment.*/

#include <stdio.h>

typedef struct
{
char a;
int b;
}A;

typedef struct
{
char a;
int b;
char c;
char d;
char e;
int f;
}B;

typedef struct
{
char x;
char y;
int z;
}C;

typedef struct
{
char x;
int z;
char y;
}D;

int main()
{
A a;
B b;
C c;
D d;
printf("Size of structure %c is %d\n",'A', sizeof(a));
printf("Size of structure %c is %d\n",'B', sizeof(b));
printf("Size of structure %c is %d\n",'C', sizeof(c));
printf("Size of structure %c is %d\n",'D', sizeof(d));

return 0;
}


the output is:

Size of structure A is 8
Size of structure B is 16
Size of structure C is 8
Size of structure D is 12


Now let me ask you a question that how come size of C is 8 while size of D is 12 while both structures contain same no. of similar elements. The answer lies in the order of elements. Further let us consider the following program:

#include <stdio.h>

typedef struct {
char c;
int j;
}t;

int main()
{
t s1;

s1.j = 4;
s1.c = 5;

printf("%p %p", &s1.j, &s1.c);

return 0;
}


and the output is:

0xbfc98f7c 0xbfc98f78


Now if you notice 0xbfc98f7c % 4 is 0 which is our alignment requirement. So, now you can see why structure C and D have different sizes.

Another point worth noting is that certain processors allocate arrays on word boundaries which are power of 2. For example, a character array requiring x bytes will actually n bytes where x is not power of 2 and n is next power of 2 which is greater than x. Such optimizations are quite common in computers and you need to be aware of it to write efficient programs because you cannot make any assumptions about behavior of processor.

1. argument:

actual argument

actual parameter (deprecated)

expression in the comma-separated list bounded by the parentheses in a function call expression, or a sequence of preprocessing tokens in the comma-separated list bounded by the parentheses in a function-like macro invocation.

Sometimes they are also called actual parameters(in function definition) but as you can see in specification ISO/IEC 9899:TC2 Section 3.3 this term is being deprecated. A function can have zero or more actual arguments and if they are more than one then each of them will be separated by a comma. These also apply to macros that is preprocessor directives when invoked like a function.

1. behavior: external appearance or action

By this term behavior the specification tries to categorize certain behavior which do not fall in other categories. For example, there are four more types of behaviors namely implementation-defined behavior, locale-specific behavior, undefined behavior and unspecified behavior. The term behavior categorizes external and those behaviors which do not fall under these four categories. External behavior includes mouse, keyboard and such.

1. implementation-defined behavior: unspecified behavior where each implementation documents how the choice is made.

EXAMPLE An example of implementation-defined behavior is the propagation of the high-order bit when a signed integer is shifted right.

When specification does not specify how a particular element of language should be implemented then programs use their logic to implement these things and sometimes it depends on hardware as well. Behavior of such elements is called implementation-defined-behavior. A full list of such behaviors is typically provided by the compiler. Consider the following example:

#include <stdio.h>

int main()
{
printf("%d %d\n", -2>>1, -1>>1);

return 0;
}


and the output is:

-1 -1

1. locale-specific behavior: behavior that depends on local conventions of nationality, culture, and language that each implementation documents

EXAMPLE An example of locale-specific behavior is whether the islower function returns true for characters other than the 26 lowercase Latin letters.

Any behavior which changes when locale of the program changes will come under this category. The character e-acute is a lowercase letter in a Latin-1 locale, but not in the “C” locale. A full list of such behaviors is typically provided by the compiler.

1. undefined behavior: behavior, upon use of a non-portable or erroneous program construct or of erroneous data, for which International Standard imposes no requirements.

NOTE: Possible undefined behavior ranges from ignoring the situation completely with unpredictable results, to behaving during translation or program execution in a documented manner characteristic of the environment (with or without the issuance of a diagnostic message), to terminating a translation or execution (with the issuance of a diagnostic message).

EXAMPLE An example of undefined behavior is the behavior on integer overflow.

We will see more examples of these undefined behaviors as we go through the specification.

It is very easy to say undefined behavior for specification but when a compiler programmer writes a compiler he cannot really say that our program is causing undefined behavior. Sometimes it will emit a warning but sometimes it will silently compile the program. However, then output of two compilers may not match. Even output of two different version of compilers may not match.

1. unspecified behavior: use of an unspecified value, or other behavior where this International Standard provides two or more possibilities and imposes no further requirements on which is chosen in any instance.

EXAMPLE An example of unspecified behavior is the order in which the arguments to a function are evaluated.

We will more of these as we progress and list of all such behaviors is given in the appendix.

It is different than undefined behavior but compiler writers are still free to to implement as they see fit.

1. bit: unit of data storage in the execution environment large enough to hold an object that may have one of two values.

NOTE It need not be possible to express the address of each individual bit of an object.

The term bit was first coined by John Wilder Tukey who also coined the term software, best known for his work on fast Fourier transform(FFT). Note that base-2 is not the most efficient way to store the numbers but it is base-e. You can read about more on it in. [Hayes] Those who are from electronics background know that transistors operate in cut-off(very low voltage, high current) and saturation(high voltage, low current). These two states consume very less power. The other two states are active and reverse-active. However, power consumption is more. Therefore, just to save power there is a trade-off between compactness and power consumption. Usually, popular compilers like gcc or gcc do not provide mechanism to sequence of bits. However, there is a compiler which allows bit-level addressing in registers if not memory. [Wegner] There are processors where data is not byte-aligned and some bit-level addressing is available as shown in [Nie] .

1. byte: addressable unit of data storage large enough to hold any member of the basic character set of the execution environment.

NOTE 1: It is possible to express the address of each individual byte of an object uniquely.

NOTE 2: A byte is composed of a contiguous sequence of bits, the number of which is implementation-defined. The least significant bit is called the low-order bit; the most significant bit is called the high-order bit.

As most of programmers will know that a byte is 8 bits. However, the term for 8 bits is octet as the term for 4 bits is nibble. This term is now so common for 8 bits that they are used interchangeably. Note that byte and char are different. byte is used to denote the storage while char is used to denote the type. We can say a 32-bit integer occupies 4 bytes but we cannot say the same in terms of char. [POSIX] defines a byte as an octet.

Considering an integer which is a multi-byte type can have its bytes individually addressed. Even though it says that numbers of bits is implementation-defined on most systems this number is governed by CHAR_BIT macro which has a value of 8.

1. character: Member of a set of elements used for the organization, control, or representation of data.

There are many character sets each capable of representing one language in this world. Sometimes they fit in a single byte while sometimes they do not and then we need more bytes to represent that kind of language.

1. single-byte character: Bit representation that fits in a byte.

2. multi-byte character: Sequence of one or more bytes representing a member of the extended character set of either the source or the execution environment.

3. wide Character: Bit representation that fits in an object of type wchar_t, capable of representing any character in the current locale.

A character is one the most primitive types of C programming language. On most machines its size is 1 byte. Even though void has also a size of 1 byte it is known as incomplete type and cannot be used to create an object.

1. constraint: Restriction, either syntactic or semantic, by which the exposition of language elements is to be interpreted.

There are certain restrictions imposed by the language through the standard which programmers must adhere to. Violation of these restrictions may lead to diagnostic messages being issued from the compiler. We will see such constraints throughout the standard.

1. correctly rounded result: Representation in the result format that is nearest in value, subject to the effective rounding mode, to what the result would be given unlimited range and precision.

This is something which is very much related to floating-point numbers. Consider two representable numbers a and b. Now there can be infinite numbers between these two. However, there will be exactly one number which will lie in between them. If this number is not representable as per floating-point specification then the rounding of this number will depend on the current rounding mode in effect. IEEE specifies four rounding modes for rounding floating-point numbers. They are: rd_near (for rounding to the nearest), rd_zero (for rounding to zero), rd_minf (for rounding to minus infinity) and rd_pinf (for rounding to plus infinity). Consider the following program which shows all four rounding modes in action.

#include <stdio.h>
#include <stdlib.h>
#include <fenv.h>

int main (int argc, char **argv)
{
float x, y, z1, z2;

x = 1.0;
y = 1.0e-20;

fesetround(FE_TONEAREST);
z1 = x - y; z2 = y - x; z1 = z1 - x; z2 = z2 + x;
printf("near, z1 = %17.10e, z2 = %17.10e \n", z1, z2);

fesetround(FE_UPWARD);
z1 = x - y; z2 = y - x; z1 = z1 - x; z2 = z2 + x;
printf("minf, z1 = %17.10e, z2 = %17.10e \n", z1, z2);

fesetround(FE_DOWNWARD);
z1 = x - y; z2 = y - x; z1 = z1 - x; z2 = z2 + x;
printf("pinf, z1 = %17.10e, z2 = %17.10e \n", z1, z2);

fesetround(FE_TOWARDZERO);
z1 = x - y; z2 = y - x; z1 = z1 - x; z2 = z2 + x;
printf("zero, z1 = %17.10e, z2 = %17.10e \n", z1, z2);

return 0;
}


and the output is:

near, z1 =  0.0000000000e+00, z2 =  0.0000000000e+00
minf, z1 =  0.0000000000e+00, z2 =  5.9604644775e-08
pinf, z1 = -5.9604644775e-08, z2 = -0.0000000000e+00
zero, z1 = -5.9604644775e-08, z2 =  5.9604644775e-08


Note that you need to link with -lm for compilation of the program, just in case.

1. diagnostic message: message belonging to an implementation-defined subset of the implementation’s message output.

These refer to compilation-time warning or error messages produced by the compiler. Since the specification does not direct any particular way in which these diagnostic messages should be generated all compiler writers are free to do whatever suits their whim. For example, earlier gcc used to print only line numbers but now they also show column numbers where errors have occurred. gcc goes one step further and produces colored output.

1. forward reference: Reference to a later subclause of International Standard that contains additional information relevant to this subclause.

1. implementation: Particular set of software, running in a particular translation environment under particular control options, that performs translation of programs for, and supports execution of functions in, a particular execution environment.

Here implementation means what we typically know as compiler. Note that it does not even say compiler. Even a C interpreter can be taken as an implementation. However, historically C has been a compiled language so we will mean implementation as compiler henceforth. Note that when you change compiler or even compiler options then you are changing the translation of code which means implementation has changed.

1. implementation limit: Restriction imposed upon programs by the implementation.

In C all types have a range or limits partially for efficiency reasons. The specification also says that a strictly conforming program will remain below the lowest limits. Now what these limits do is that increase the portability of program just like being the lowest common denominator.

1. memory location: Either an object of scalar type, or a maximal sequence of adjacent bit-fields all having nonzero width.

1. object: region of data storage in the execution environment, the contents of which can represent values. When referenced, an object may be interpreted as having a particular type.

These are what we know as variables in common developer language. The region of data storage occupied by an object will be contiguous number of bytes. Note that the term object used in ISO standard of C has nothing to do with object in object-oriented programming paradigm.

1. parameter: Also known as formal parameter or formal argument (deprecated). Object declared as part of a function declaration or definition that acquires a value on entry to the function, or an identifier from the comma-separated list bounded by the parentheses immediately following the macro name in a function-like macro definition.

For example:

#define FUNC(X, Y) //two parameters

int f(int x);  //one parameter
void y(int y, int z); //two parameters

1. recommended practice: Specification that is strongly recommended as being in keeping with the intent of the standard, but that may be impractical for some implementations.

1. runtime-constraint: Requirement on a program when calling a library function.

1. value: Precise meaning of the contents of an object when interpreted as having a specific type.

This at times will depend on the type of machine. For example, a union containing an integer and two characters will have different values for characters on big-endian and little-endian machines. But on the same machine the meaning must be accurate for contents of an object.

1. implementation-defined value: Unspecified value where each implementation documents how the choice is made.

There are certain values defined by the implementation. For example, in the table for defining numerical limits in chapter 3 specification defines INT_MAX in a fashion as if integers are 16-bit. But all modern 32-bit compilers do not follow this limit and they treat integer as 32-bit entity.

1. indeterminate value: Either an unspecified value or a trap representation.

Consider a variable declaration int i;. What value does this integer i contain? We do not know. The value contained is unspecified as per specification nomenclature. Developers also know this as garbage value. Behavior caused by usage of such variables will be known as unspecified behavior. However, behavior caused by a trap representation will be undefined behavior.

1. unspecified value: Valid value of the relevant type where International Standard imposes no requirements on which value is chosen in any instance.

NOTE: An unspecified value cannot be a trap representation.

As we have seen in last example the value of the variable will be unspecified value.

1. trap representation: An object representation that need not represent a value of the object type.

2. perform a trap: Interrupt execution of the program such that no further operations are performed.

NOTE In this International Standard, when the word “trap” is not immediately followed by “representation”, this is the intended usage.

3. $$\lceil x\rceil$$: ceiling of x: the least integer greater than or equal to x.

4. $$\lfloor x\rfloor$$: floor of x: the greatest integer less than or equal to x.

Hayes
1. Hayes. Third base. American Scientist, 89(6):490–494, 2001.

Wegner

J. Wagner and R. Leupers. C compiler design for an industrial network processor. In Proceedings of The Workshop on Languages, Compilers, and Tools for Embedded Systems (LCTES 2001), pages 155–164, 2001.

Nie

X. Nie, L. Gazsi, F. Engel, and G. Fettweis. A new network processor architecture for high-speed communications. In IEEE Workshop on Signal Processing Systems (SiPS’99), 1999.

POSIX

ISO. ISO/IEC FDIS 9945:2008 Information technology - Portable Operating System Interface (POSIX(R)). ISO, 2008.

You can also read the conformance part of this chapter which is chapter 3 in specification.