C Basics

● In this course, we’ll use the C programming language.

○ Specifically, we’ll use the C99 standard of the C language.

● We’ll assume you already know C++. Since C is a subset of C++, this means
making the jump from C++ to C shouldn’t be too hard. Still, there are some key
differences between the way things are done in C++ and the way they’re done in
C, so we’ll start the course by covering some basics of the C language.
○ If you don’t already know C++, you should use a resource like the C
tutorial on cprogramming.com to provide a more in-depth look at some of
the C features we’ll gloss over here (like loops, functions, if/else
statements, etc.).

● Importantly, for this course, we’ll use the GCC C compiler, which is installed on
all of the ENGR machines. The GCC C compiler (gcc) works very much like the
GCC C++ compiler (g++), which you’ve likely used. For example, here’s how to
use the GCC C compiler to compile (under the C99 standard) a single C file
called main.c to produce an executable file named main:

gcc --std=c99 -o main main.c

● Let’s start by reviewing some essentials.

C Program structure
● The basic C program template is just like in C++. Specifically, every program
needs a function called main(), which serves as the entrypoint into the program.
This file typically returns an integer and accepts arguments describing the
number of arguments passed on the command line (argc) and providing the
string values of those command line arguments (argv; we’ll talk more later about
how strings are represented in C):

int main(int argc, char** argv) {

return 0;
}

● C programs will typically also contain one or more include statements at the top
of the file. Like in C++, these incorporate external code into the current file.
 Unlike in C++, we need to specify the complete name of the file being included
(including the file extension) in all include statements.
○ The standard file extension for header files in C is .h.

● For example to include functionality for writing text to and reading it from the
console (or a file), we would include the (built-in) standard I/O header:

#include <stdio.h>

Printing text to stdout with printf()

● One of the simplest things we’ll want to be able to do in a C program is to print
text to the terminal (i.e. to the standard output stream, or stdout). This is done
differently in C than in C++. Specifically, in C, we use the function printf() to
print text to stdout:

printf("This is a string I’m printing to stdout.\n");

● In the example above, we’re simply printing a constant string to stdout.

Importantly, note that we explicitly include the newline (\n) character in this string
to include a line break in the text printed to stdout. In general, we can print as
many line breaks as we want by including multiple newline characters in the
string passed to printf().

● What if we want to do more than print a constant string? For example, what if we
want to print the values held in some of our program’s variables?

● We can do this by passing a format string and accompanying arguments to

printf().

● You can think of a format string as a template for the text to be printed. In
particular, a format string will contain placeholders (known as format specifiers)
into which specific values will later be inserted.

● Each format specifier is designated by a % character, followed by at least one

other character describing the data that will eventually be plugged in.

● For example, if we had an integer variable (i.e. an int) called x whose value we
wanted to print, we could include a %d specifier (i.e. placeholder) in our format
string and then, importantly, pass x as an additional argument to printf():
 int x = 8;
printf("This is the value of x: %d\n", x);

● There are lots of different kinds of format specifiers we can use:

○ %d – indicates an int, to be printed as a signed decimal number
○ %f – indicates a double, to be printed in fixed-point notation (e.g.
3.1415…)
■ float arguments are cast as double
○ %c – indicates a char, to be printed as a readable character
○ %s – indicates a null-terminated string
○ Lots more…

● We can also print multiple values by inserting multiple format specifiers. Then,
the i’th argument to printf() (after the format string) is plugged into the i’th
format specifier, e.g.:

char* name = "Luke Skywalker";

double gpa = 3.75;
printf("%s's GPA is %f\n", name, gpa);

If statements
● For program control, C has several features that work basically the same way as
they do in C++. If statements are one of these:

if (a == 0) {
/* Do something. */
} else if (b != 0) {
/* Do something different. */
} else {
/* Do a third thing altogether. */
}

Loops
● Loops also work pretty much the same way in C as they do in C++. C has for
loops:

int i;
 for (i = 0; i < 32; i++) {
/* Do something 32 times. */
}

● C also has while loops:

while (i != 16) {
/* Do something repeatedly until i is 16. */
}

● Like C++, C also has do/while loops, though these are less commonly used.

Functions
● C also has functions that work mostly like they do in C++ (except they’re all at file
scope: no class methods):

#include <stdio.h>

void foo(int x) {
printf("foo was passed this argument: %d\n", x);
}

int main(int argc, char** argv) {

foo(2);
}

● Just like C++, you can prototype a function first (e.g. in a header file) and define it
later:

#include <stdio.h>

/* This could be in a separate .h file too */

void foo(int);

int main(int argc, char** argv) {

foo(2);
}

/* This could be in a separate library .c file */

void foo(int x) {
 printf("foo got %d\n", x);
}

● C has no reference types, unlike C++! That means functions are all pass-by-
value.
○ In other words, a copy of each argument is passed to the function as a
parameter. This can be a problem if we’re passing big structures around.
We’ll talk more about this later.

● The main difference between C and C++ is that C is procedural, not object
oriented like C++.
○ This means that C has no classes or class methods. Instead, all
operations on a given data entity are performed by passing that entity to a
function.
○ As we’ll discuss later, structured data is represented in C with struct.
○ For example, we might do something like this in C++:

Student s = new Student("Luke Skywalker");

s.print();

○ In C, we’d do something like this:

struct student s = {.name = "Luke Skywalker"};

print_student(s);

Structures
● If we want to represent structured data in C, we typically use a struct type. For
C++ programmers, a struct is much like a class with only public data members
and no class methods.

● For example, here’s what a definition of a struct for representing individual

students might look like:

struct student {
char* name;
int id;
float gpa;
};
 ● We can initialize a new struct with what’s known as a designated initializer:

struct student s =
{.name = "Luke Skywalker", .gpa = 4.0};

○ Note that, like in the example here, we don’t need to initialize all fields of a
struct with a designated initializer. Uninitialized fields will be zeroed.

● We can access or update a field in a struct using the . operator (assuming we

have a struct value, like s here, and not a struct pointer; more on this in a bit):

s.id = 933111111;
printf("%d\n", s.id);

Pointers
● A pointer is a variable whose value is a memory address. Every pointer points to
data of a specific type. For example, we can initialize a pointer to a single integer
value like this:

int n = 20;
int* n_ptr = &n;
● In the example above, n is a normal integer value, and n_ptr is a pointer whose
value is the memory address where n is stored (i.e. &n). In other words, n_ptr
points to n.

● This becomes even more clear if we print out the values associated with these
two variables:

printf("n: %d\n", n);

printf("n_ptr: %p\n", n_ptr);

○ Note here that we print the pointer value using a %p format specifier. This
is used specifically for printing pointer values.

● The output of these two lines will look something like this:

n: 20
n_ptr: 0x7ffee5032958
● We can see here more explicitly that n represents the value of n, and n_ptr
represents the memory address where n is stored.

● If we want to obtain the underlying value that’s stored at a particular memory

address, we must dereference the pointer using the * operator:

printf("*n_ptr: %d\n", *n_ptr); /* Prints 20. */

● We can also dereference a pointer to update the value stored at the

corresponding memory address, e.g.:

*n_ptr = 8;
printf("n: %d\n", n);
printf("n_ptr: %p\n", n_ptr);
printf("*n_ptr: %d\n", *n_ptr);

● This will print something like this:

n: 8
n_ptr: 0x7ffee5032958
*n_ptr: 8
● Note here that the underlying value has changed, but that value is stored at the
same memory address (i.e. 0x7ffee5032958).

● Importantly, C, unlike C++, has no pass-by-reference, but we can achieve the

same thing with pointers:

/* Takes an *address* of an int */

void make_it_32(int* a) {
*a = 32;
}

int main() {
int i = 6;
make_it_32(&i); /* Pass the *address* of i. */
printf("%d\n", i); /* Prints 32. */
}
Void pointers (void*)
● In C, you’ll often run into something called a void pointer. This is a pointer
represented by the type void*.

● A void pointer is a generic pointer. In other words, it can point to data of any data
type. For example, we could make a void pointer that pointed to an integer:

int n = 32;
void* v_ptr = &n;

● If we tried to use any different pointer type, e.g. float*, to point to an integer
value, the compiler would give us a warning. However, using a void* works just
fine.

● We can use a void pointer to point to values of any other type, too, though:

float pi = 3.1415;
struct student s = { .name = "Luke Skywalker" };
v_ptr = &pi; /* Totally valid. */
v_ptr = &s; /* Totally valid. */

● Importantly, because there is no type information built into a void pointer, it can’t
be dereferenced directly. In other words, we can’t do this:

struct student s = { .name = "Luke Skywalker" };

void* v_ptr = &s;
printf("%s\n", v_ptr->name); /* Compile-time error: can’t
dereference void pointer */

● Instead, we must either assign the void pointer back to a pointer variable of the
correct type or else cast it to the correct type and then dereference it.

● For example, here’s how we could assign the void pointer v_ptr we created just
above back to a pointer variable of the correct type (i.e. struct student*) in
order to dereference it:

struct student* s_ptr = v_ptr;

printf("%s\n", s_ptr->name);
 ● And here’s how we could just cast back to the correct pointer type first to
dereference the void pointer:

printf("%s\n", ((struct student*)v_ptr)->name);

● You’ll often see void pointers used in data structures implemented in C, since
they allow the data structures to contain data of any type while remaining type
agnostic.

Program memory: the call stack vs. the heap

● A running C program (or a program in any language, for that matter) has two
separate areas of memory in which it can store data, the call stack and the heap.
○ The call stack is a small, limited-size chunk of memory from the larger
blob of system memory. Among other things, the values of variables
declared in a program’s functions are stored on the call stack.
■ The call stack is small: usually at most 8kb.
○ The heap comprises essentially all the rest of system memory.
■ A program must specifically request to allocate memory from the
heap. We’ll see how to do this in C in just a second.
■ The heap is huge compared to the call stack.
● Here’s what this might look like for a simple program:

int main() {
int i = 16;
int* i_ptr = &i;
}
 All System

The

Th
e
i =
i_p

He
ap
...
...

Allocating memory on the heap: malloc()

● Again, the call stack is quite small (e.g. only 2048 4-byte integers fit on an 8kb
stack), so we’ll need some way to work with bigger chunks of memory. The heap
allows us to do this.

● When we want to use memory from the heap, though, we need to explicitly ask to
have it allocated. In C++, we did this using the new operator. Unfortunately, new
is not available in C. Instead, we’ll use the malloc() function.

● The malloc() function requires us to #include <stdlib.h>. It takes as its

argument the number of bytes to be allocated from the heap and returns the
allocated bytes as a void pointer, which can be cast as a pointer of any type:

void* allocated_memory = malloc(NUMBER_OF_BYTES);

● Importantly, malloc() allocates a contiguous block of memory. This means

that the memory is allocated as a single, continuous chunk, rather than
potentially in multiple pieces. This, for example, makes it easy to use malloc()
to allocate an array.

● Again, memory allocated using malloc() lives on the heap. This is known as
dynamically allocated memory. It is allocated at runtime.
 ○ Memory allocated on the call stack, on the other hand, is known as
statically allocated memory. It is allocated at compile time.

● You may be wondering how one might figure out how many bytes to allocate
when using malloc(). The sizeof() operator is very useful for this.
sizeof() returns the size, in bytes, of a given variable or data type. For
example, if we wanted to know the number of bytes required for a single integer,
we could use sizeof() like so:

sizeof(int)

● Thus, to let malloc() know how many bytes to allocate from the heap, we can
make use of sizeof(), e.g.:

int* n_ptr = malloc(sizeof(int));

*n_ptr = 13;

● This would look somethingAll

like this within system memory:

Call
13 ...
n_p

● To allocate an array, we can simply multiply the value returned by sizeof() by

the desired size of the array. For example, here’s how we could allocate an
array of 1000 integers on the heap:

int* array = malloc(1000 * sizeof(int));

● Once an array is allocated on the heap, it is used just like any other array:
 int i;
for (i = 0; i < 1000; i++) {
array[i] = i;
}

malloc() and struct

● We can use malloc() with any struct just like we do for C’s standard types to
allocate a struct on the heap, e.g.:

struct student s* = malloc(sizeof(struct student));

● This would look something

All like this in system memory:
System

Call
s
name
= ...
standing

● When we have a pointer to a struct, we need to dereference the pointer to

access the struct’s fields:

(*s).name = "Luke Skywalker";

(*s).gpa = 4.0;

● That’s so much typing just to get to one field! Luckily C has the -> operator that
both dereferences a struct pointer and accesses one of its fields:

s->name = "Luke Skywalker";

 s->gpa = 4.0;

● We allocate an array of structs just like an array of any other type, e.g.:

struct student* students =

malloc(1000 * sizeof(struct student));

Freeing malloc()’ed memory

● It’s incredibly important that we ALWAYS free all of the memory allocated using
malloc(). If we don’t our program will have what’s called a memory leak.
Memory leaks can cause unpredictable, catastrophic failure of a program, so it’s
best to avoid them.

● Memory allocated with malloc() is always freed using free():

int* array = malloc(1000 * sizeof(int));

...
free(array);

● A good rule of thumb is this: For every call to malloc() you should have a
corresponding call to free().

● free() works the same regardless of what kind of memory is being freed. In
other words, whether a chunk of allocated memory holds a single value, an array,
a struct, or whatever, we simply pass a pointer to that memory to free(), and it
will be released.

Using valgrind to debug memory issues

● Memory issues, like leaks, are sometimes the most insidious ones to diagnose
and debug. Fortunately, there are some useful tools available to us to help
debug memory issues. valgrind is one of these tools.

● valgrind works both by inserting extra instructions into a program being executed
and by replacing the standard system memory allocator with its own
implementation. The combination of these modifications to the program being
executed allows valgrind to closely monitor that program’s memory usage and to
detect several kinds of memory error.
 ● To make valgrind most useful, it is best to include debug flags in the program to
be debugged. We can do this by passing the -g option to gcc:

gcc --std=c99 -g prog.c -o prog

● Once the program is compiled into an executable, we can run that program
through valgrind on the command line by simply adding the valgrind
command before our normal command-line invocation of our program:

valgrind ./prog [args to prog]

● Valgrind will then execute your program, and, among other things, it will detect
memory leaks. If you have any leaks, it will let you know that some memory was
“lost”.

● To dig deeper into where memory was lost, pass the --leak-check=full
option to the valgrind command:

valgrind --leak-check=full ./prog [args]

● If debug flags are compiled into the program being executed, the --leak-
check=full option will tell valgrind to print a report with the line numbers of the
malloc() calls for the memory that was lost. This is extremely helpful for
tracking down allocated memory you may have forgotten to free.

Strings in C
● In C, there is no special type for representing character strings, unlike in C++,
where we can use the std::string class.

● Instead, character strings are represented in C as arrays of char values,

typically as char*.

● Importantly, all strings in C use a special character value to indicate the end of
the string. This value is known as the null character and is represented as the
character '\0' (i.e. a backslash-escaped zero character).

● Thus, in addition to the normal characters within a string, every valid C string
contains this extra null character. For example in memory, the string “cat” would
look like this in C:
 0 1 2 3

c a t \0

● In other words, in C, the string “cat” would be represented as an array of 4

characters: 'c', 'a', 't', and '\0'.

● The null character is extremely important in a C string. This is how C knows

where the string ends. Functions that use strings—including printf() but also
including several other functions, some of which we’ll see in a second—rely on
the presence of the null character to know when to stop processing the string.
○ For example, strlen(), which simply returns the number of characters in
a string, essentially counts characters by starting at the beginning of the
string and incrementing until it finds a null character.

● Thus, if we are manually building a string, we must make sure to include a null
character at the end to terminate it.

● Similarly, if we are allocating memory in which to store a string, we need to make

sure to allocate enough space to hold all of the normal characters in the string
plus the terminating null character.

● For example, the following line allocates enough space for a string with 63
normal characters and a terminating null character:

char* str = malloc(64 * sizeof(char));

● C allows us to initialize constant strings as you might expect, e.g.:

char* name = "Leia Organa";

● However, when we create a constant string this way, it is important to remember

that the string itself is read-only. In other words, a string created this way cannot
be modified. For example, this would be an illegal operation on the above string:

name[0] = 'l';
● Unfortunately, most C compilers will not issue an error or warning if you try to
modify a read-only string like this. Instead, the program will simply crash at
runtime (often with a very cryptic error message).

● Thus, it is typically best to explicitly mark constant strings with the const
keyword. By doing this, we ensure that the compiler will throw an error if we try
to modify the string. For example, this will result in a clear compiler error:

const char* name = "Leia Organa";

name[0] = 'l'; /* This will be a compile-time error. */

● Finally, there are a lot of utility functions available in C to help us work with
strings. Many of these will require you to use the following include statement:

#include <string.h>

● One very useful string utility function is strlen(), mentioned above. This
simply returns the number of characters in a string, not including the terminating
null character. For example, strlen("cat") will return 3 not 4.

● Another useful string utility function from string.h is strncpy(), which will copy
up to a specified number of characters from one string to another. Here is a
typical use that combines both strncpy() and strlen():

char* name = "Leia Organa";

int n = strlen(name);
char* copy = malloc((n + 1) * sizeof(char));
strncpy(copy, name, n + 1);

● The third argument to strncpy(), which above is n + 1, limits the number of

characters that will be copied into the destination string (i.e. copy above). This
argument must provide enough space for the terminating null character to be
copied, which is why we use n + 1 above instead of n.
○ Our use of malloc() above also accounts for the trailing null character
by allocating space for n + 1 characters.

● Another useful string utility function is snprintf(), which is included via

stdio.h. snprintf() provides printf()-like functionality for “printing”
content into a string, up to a specified number of characters.
 ● For example, here, we print a representation of a complete student struct into a
single string using snprintf():

struct student s = {.name="Rey", .id=933222222, .gpa=3.8};

int maxlen = 128;
char* str = malloc(maxlen * sizeof(char));
snprintf(str, maxlen, "%s %d %f", s.name, s.id, s.gpa);

● The string stored in str here will look like this: "Rey 933222222 3.8".

● Importantly, the maxlen argument we pass here to snprintf() limits the

number of characters that can be “printed” into the string str. In particular, this
will result in at most maxlen - 1 normal characters being “printed” to leave
room for the terminating null character, which snprintf() automatically inserts.

● Note that there are versions of strncpy() and snprintf() (specifically

strcpy() and sprintf()) that do not require an argument to be passed to
specify the maximum number of characters to store in the destination string. You
are STRONGLY DISCOURAGED from using these functions, since they are
highly vulnerable to a particularly insidious form of attack.

● We’ve only explored a very small subset of the string utility functions available in
C. Check out what’s in string.h to see more.

Function pointers (advanced C)

● Function pointers are an advanced feature of the C language that allows us to
store the memory address of a function in a variable and to use that memory
address to call the function being pointed to.

● To understand why function pointers are useful, it’s helpful to know that the
primary way function pointers are used is to allow us to pass functions as
arguments to other functions.
○ If you’re a JavaScript programmer, you probably just felt a mild, warm,
fuzzy feeling thinking about passing functions as arguments to other
functions.
● Why would we want to pass a function as an argument to another function? To
see why, let’s explore a common situation in which this is used.

● Imagine you’re writing a C function to sort an array. You’d like this function to be
useful in a wide variety of situations. Specifically, instead of limiting this function
to sorting arrays of only one type (e.g. only int arrays, only double arrays,
etc.), you’d like to implement it so that it can potentially sort arrays of any type.

● Your first step here might be to specify the function so that the array to be sorted
is passed as a void pointer. For example, your sorting function’s prototype might
look like this:

void sort(void* arr, int n);

● So far, so good. A function prototype like this will allow us to pass an array of
any kind of data into our sort() function (ignoring for now the fact that we don’t
know how big each element of the array is, i.e. how many bytes it occupies).
● One major problem remains, though, which is, without knowing the type of data
it’s sorting, how will our sort() function be able to compare the values in the
array? In other words, how will it figure out how to order the data, so it can move
smaller values to the front of the array and larger ones to the end?

● This is where function pointers come into play. Again, remember that a key
feature of function pointers is that they allow us to pass functions as arguments
to other functions.

● In particular, while our sort() function doesn’t know the type of the data it's
sorting and thus doesn’t know how to compare it, the function calling our sort()
function and passing data into it probably does know these things.

● Thus, our sort() function could require its calling function to provide, in addition
to the array to be sorted, a function to compare elements of that array to each
other to determine which ones are bigger and which ones are smaller.

● This can be accomplished by adding a third argument to our sort() function’s

prototype. This argument will be a function pointer argument, and it will be used
to require the calling function to provide a function for comparing elements in the
array to be sorted.
● To use a function pointer, we must know the exact prototype of the function we
want to point to. In the context of our sort() function, we’ll want the calling
function to provide a function that compares two values from the array to be
sorted and returns a value indicating which is bigger.

● In this specific situation, where we’re sorting data in an array represented as a

void pointer, the comparison function could take two void pointers as arguments,
where the two void pointers point to different elements in the array to be
compared with each other), and it could return an integer value indicating which
one is bigger than the other (e.g. it could return 0 if one is bigger or 1 if the other
is). In other words, the prototype of the comparison function could look like this:

int cmp(void* a, void* b);

● Let’s include into our sort() function’s prototype an additional function pointer
argument matching the comparison function prototype just above. This is what
it’d look like:

void sort(void* arr, int n, int (*cmp)(void* a, void* b));

● You should be able to see here how the function pointer specification
corresponds to the comparison function prototype above. Specifically it has an
int return type and takes two void* arguments, a and b. The only difference is
that now the function name is represented using pointer syntax, i.e. as (*cmp).

● How can we use this? Let’s first take a look at how our sort() function will be
called. We’ll explore a simple case, where the calling function wants to sort an
array of integers.

● First, the calling function will need access to a function for comparing integers.
This function will have to match the prototype specified by the function pointer
argument to our sort() function. This means it will have to accept the pointers
to the integers to be compared, specifically represented as void pointers. To
perform the comparison, it can simply cast the void pointers back to integer
pointers and then dereference:

int compare_ints(void* a, void* b) {

int* ai = a, * bi = b; /* Cast void* back to int*. */
 if (*ai < *bi) {
return 0;
} else {
return 1;
}
}

● Here, we assume that our sort() function expects the comparison function to
return 0 if a < b and 1 otherwise.

● Now, when the calling function invokes our sort() function, it can pass a
pointer to compare_ints as the third argument:

sort(array_of_ints, number_of_ints, compare_ints);

● How do we use the comparison function then, within our sort() function?
Whenever we need to compare two values from the array being sorted, we can
just call cmp() (which is the name given to the function pointer argument), e.g.:

if (cmp(arr[i], arr[j]) == 0) {
/* Put arr[i] before arr[j] in the sorted array. */
} else {
/* Put arr[i] after arr[j] in the sorted array. */
}

● In this way, the calling function can simply provide a comparison function for
whatever data it wants to sort, and our sort() function can remain agnostic
about what kind of data it’s actually operating on. This kind of approach will be
extremely useful in allowing us to implement completely general-purpose data
structures in C.

● Indeed, this kind of approach is common practice for widely-used C utilities, such
as the qsort() function, which does almost exactly what our sort() function
here does using a similar approach.