311 (5) C

Page 1:

• Course: CMPSC 311 - Introduction to Systems Programming

• Professor: Sencun Zhu

• Slides by: Professor Patrick McDaniel and Professor Abutalib Aghayev

Page 2:

• Files in C: foo.c, foo.h, bar.c

• C workflow: Editing, linking, executing, debugging, profiling

• Types of libraries: Statically linked libraries (libZ.a), Shared libraries (libc.so)

• Object files: bar.o, foo.o

Page 3:

• Defining a function in C

• Example: sumTo function

  • Parameters: int max

  • Variables: int i, int sum

  • Loop: for (i=1; i<=max; i++)

  • Statement: sum += i

  • Return: return sum

Page 4:

• Conversion from C to machine code

• Example: dosum function

• C source file: dosum.c

• C compiler: gcc -S

• Assembly source file: dosum.s

• Assembler: as

• Machine code: dosum.o

Page 5:

• C compilers generate object ".o" files directly

• Object code is re-locatable machine code

• Object code generally cannot be executed without manipulation (e.g., via a linker)

Page 6:

• Anatomy of a C program

• Includes: #include <stdio.h>

• Functions: main() and myfunc()

• Example program

Page 7:

• Running a C program

• Compiling and linking: gcc -g -Wall main.c -o main

• Executing the program: ./main

Page 8:

• Running a program in UNIX

• Search path: PATH environment variable

• Adding to search path: export PATH=$PATH:/new/path

Page 9:

• Multi-file C programs

• Example: dosum function in dosum.c

• Example: main function in sumnum.c

• Prototyping functions

Page 10:

• Multi-file C programs

• Including standard libraries: #include <stdio.h>

Page 11:

• Compiling multi-file programs

• Linking multiple object files to produce an executable

• Linking standard libraries

Page 12:

• Object files in C

• Object files contain machine code produced by the compiler

• Object files might contain references to external symbols

• Linking resolves external symbols

Page 13:

• Similarities between C and Java

• Syntax, types, type-casting, expressions, operators, scope, comments

Page 14:

• Primitive types in C

• Integer types: char, int, short, long

• Floating point types: float, double

• Type modifiers: signed, unsigned

Page 15:

• C99 extended integer types

• Solve the conundrum of "how big is a long int?"

• Example using stdint.h library

Page 16:

• Variables in C are similar to Java

  • Variables must be declared at the start of a function or block (not required since C99)

  • Variables need not be initialized before use (gcc -Wall will warn)

  • It is always recommended to initialize variables before use

• Example code:

  #include <stdio.h>
  int main(void) {
    int x, y = 5;  // note x is uninitialized!
    long z = x+y;
    printf("z is '%ld'\n", z); // what’s printed?
    {
      int y = 10;
      printf("y is '%d'\n", y);
    }
    int w = 20;  // ok in c99
    printf("y is '%d' , w is '%d'\n", y, w);
    return 0;
  }

Page 17:

• The const qualifier in C indicates that a variable's value cannot change

• The compiler will issue an error if you try to violate this

• Example code:

  #include <stdio.h>
  int main(void) {
    const double MAX_GPA = 4.0;
    printf("MAX_GPA: %g\n" , MAX_GPA);
    MAX_GPA = 5.0;  // illegal!
    return 0;
  }

Page 18:

• Loops in C are similar to Java

  • For loops cannot declare variables in the loop header (changed in C99)

  • If/else, while, and do while loops are available

  • C does not have a boolean type (changed in C99 with #include <stdbool.h>)

  • Any type can be used in conditionals, where 0 means false and everything else means true

• Example code:

  int i;
  for (i=0; i < 100; i++) {
    if (i % 10 == 0) {
      printf("i: %d\n" , i);
    }
  }

Page 19:

• Pointers in C allow you to store memory addresses of variables

• Key concepts:

  • Taking the address of a variable: &

  • Dereferencing a pointer: *

  • Aliasing: *ip is an alias for i

• Example code:

  #include <stdio.h>
  int main(void) {
    int i = 5;
    int *ip = &i;
    printf("%d\n" , i);
    printf("%p\n", ip);
    *ip = 42;
    printf("%d\n", i);
    printf("%d\n", *ip);
  }

Page 20:

• C always passes arguments by value

• Pointers allow you to pass arguments by reference

• Example code:

  void add_pbv(int c) {
    c += 10;
    printf("pbv c: %d\n", c);
  }
  void add_pbr(int *c) {
    *c += 10;
    printf("pbr *c: %d\n", *c);
  }
  int main(void) {
    int x = 1;
    printf("x: %d\n", x);
    add_pbv(x);
    printf("x: %d\n", x);
    add_pbr(&x);
    printf("x: %d\n", x);
    return 0;
  }

Page 21:

• C passes arguments by value

• If the callee modifies an argument, the caller's copy isn't modified

• Example code:

  void swap(int a, int b) {
    int tmp = a;
    a = b;
    b = tmp;
  }
  int main(void) {
    int a = 42, b = -7;
    swap(a, b);
    printf("a: %d, b: %d\n", a, b);
    return 0;
  }

Page 30:

• Pointers can be used to pass arguments by reference in C

• The callee still receives a copy of the argument, but it is a pointer to the variable in the scope of the caller

• Example code:

  void swap(int *a, int *b) {
    int tmp = *a;
    *a = *b;
    *b = tmp;
  }
  int main(void) {
    int a = 42, b = -7;
    swap(&a, &b);
    printf("a: %d, b: %d\n", a, b);
    return 0;
  }

• The key to C (and languages like it) is getting good at using pointers.

Page 31:

• Pass-by-reference OS kernel [protected] stack main a 42 b -7

• void swap(int *a, int *b) { int tmp = *a; *a = *b; *b = tmp; }

• int main(void) { int a = 42, b = -7; swap(&a, &b); printf("a: %d, b: %d
  ", a, b); return 0; }

• CMPSC 311 - Introduction to Systems Programming

Page 32:

• Pass-by-reference OS kernel [protected] stack main a 42 b -7

• void swap(int *a, int *b) { int tmp = *a; *a = *b; *b = tmp; }

• int main(void) { int a = 42, b = -7; swap(&a, &b); printf("a: %d, b: %d
  ", a, b); return 0; }

• swap a ? b ?

• tmp ?

• CMPSC 311 - Introduction to Systems Programming

Page 33:

• Pass-by-reference OS kernel [protected] stack main a 42 b -7

• void swap(int *a, int *b) { int tmp = *a; *a = *b; *b = tmp; }

• int main(void) { int a = 42, b = -7; swap(&a, &b); printf("a: %d, b: %d
  ", a, b); return 0; }

• swap a b tmp ?

• CMPSC 311 - Introduction to Systems Programming

Page 34:

• Pass-by-reference OS kernel [protected] stack main a 42 b -7

• void swap(int *a, int *b) { int tmp = *a; *a = *b; *b = tmp; }

• int main(void) { int a = 42, b = -7; swap(&a, &b); printf("a: %d, b: %d
  ", a, b); return 0; }

• swap a b tmp 42

• CMPSC 311 - Introduction to Systems Programming

Page 35:

• Pass-by-reference OS kernel [protected] stack main a -7 b -7

• void swap(int *a, int *b) { int tmp = *a; *a = *b; *b = tmp; }

• int main(void) { int a = 42, b = -7; swap(&a, &b); printf("a: %d, b: %d
  ", a, b); return 0; }

• swap a b tmp 42

• CMPSC 311 - Introduction to Systems Programming

Page 36:

• Pass-by-reference OS kernel [protected] stack main a -7 b 42

• void swap(int *a, int *b) { int tmp = *a; *a = *b; *b = tmp; }

• int main(void) { int a = 42, b = -7; swap(&a, &b); printf("a: %d, b: %d
  ", a, b); return 0; }

• swap a b tmp 42

• CMPSC 311 - Introduction to Systems Programming

Page 37:

• Pass-by-reference OS kernel [protected] stack main a -7 b 42

• void swap(int *a, int *b) { int tmp = *a; *a = *b; *b = tmp; }

• int main(void) { int a = 42, b = -7; swap(&a, &b); printf("a: %d, b: %d
  ", a, b); return 0; }

• CMPSC 311 - Introduction to Systems Programming

Page 38:

• Pass-by-reference OS kernel [protected] stack main a -7 b 42

• void swap(int *a, int *b) { int tmp = *a; *a = *b; *b = tmp; }

• int main(void) { int a = 42, b = -7; swap(&a, &b); printf("a: %d, b: %d
  ", a, b); return 0; }

• CMPSC 311 - Introduction to Systems Programming

Page 39:

• Very different than Java

• arrays

• just a bare, contiguous block of memory of the correct size

• array of 6 integers requires 6 x 4 bytes = 24 bytes of memory

• arrays have no methods, do not know their own length (no bounds checking)

• C doesn’t stop you from overstepping the end of an array!

• many, many security bugs come from this (buffer overflow)

• CMPSC 311 - Introduction to Systems Programming

Page 40:

• Very different than Java

• arrays

• just a bare, contiguous block of memory of the correct size

• array of 6 integers requires 6 x 4 bytes = 24 bytes of memory

• arrays have no methods, do not know their own length (no bounds checking)

• C doesn’t stop you from overstepping the end of an array!

• many, many security bugs come from this (buffer overflow)

• X[7] = 45; // Legal C, but can cause memory fault! ! ! !

• CMPSC 311 - Introduction to Systems Programming

Page 41:

• Very different than Java

• strings

• array of char

• terminated by the NULL character ‘\0’

• are not objects, have no methods; string.h has helpful utilities (see strings lecture coming soon!)

• h e l l o
  \0

• char *x = ”hello
  ”;

• x

• CMPSC 311 - Introduction to Systems Programming

Page 42:

• Very different than Java

• errors and exceptions

• C has no exceptions (no try / catch)

• errors are returned as integer error codes from functions

• sometimes makes error handling ugly and inelegant

• some support from OS using signals (end of semester)

• crashes

• if you do something bad, you’ll end up spraying bytes around memory

• hopefully causing a “segmentation fault” and crash

• objects

• there aren’t any; struct is closest feature (set of fields)

• CMPSC 311 - Introduction to Systems Programming

Page 43:

• Very different than Java

• memory management

• there is no garbage collector

• anything you allocate you have to free (memory leaks)

• local variables are allocated off of the stack

• freed when you return from the function

• global and static variables are allocated in a data segment

• are freed when your program exits

• you can allocate memory in the heap segment using malloc()

• you must free malloc’ed memory with free()

• failing to free is a leak, double-freeing is an error (hopefully crash)

• CMPSC 311 - Introduction to Systems Programming

Page 44:

• Very different than Java

• console I/O

• C library (libc) has portable routines for reading/writing, e.g., scanf() , printf()

• file I/O

• C library has portable routines for reading/writing

• fopen() , fread() , fwrite() , fclose() , etc.

• does buffering by default, is blocking by default

• OS provides system calls

• we’ll be using these: more control over buffering, blocking

• Low level binary reads and writes, e.g., read() , write() , open() , close()

• CMPSC 311 - Introduction to Systems Programming

Page 45:

• Very different than Java

• network I/O

• C standard library has no notion of network I/O

• OS provides (somewhat portable) routines

• lots of complexity lies here

• errors: network can fail

• performance: network can be slow

• concurrency: servers speak to thousands of clients simultaneously

• Note: most of these topics will be covered in detail over the semester.

• CMPSC 311 - Introduction to Systems Programming

Page 46:

• Very different than Java

• Libraries you can count on

• C has very few compared to most other languages

• no built-in trees, hash tables, linked lists, sort , etc.

• you have to write many things on your own

• particularly data structures

• error prone, tedious, hard to build efficiently and portably

• less productive language than Java, C++, python, or others

• CMPSC 311 - Introduction to Systems Programming

Page 47:

• Problem: ordering

• Don’t call a function that hasn’t been declared yet:

• #include <stdio.h>

• int main(void) { printf("sumTo(5) is: %d
  ", sumTo(5)); return 0; }

• // sum integers from 1 to max

• int sumTo(int max) { int i, sum = 0; for (i=1; i<=max; i++) { sum += i; } return sum; }

• CMPSC 311 - Introduction to Systems Programming

Page 48:

• Problem: ordering

• Solution 1: reverse order of definition

• #include <stdio.h>

Page 49:

Problem: ordering

• Solution 2: provide function declaration

  • Teaches the compiler the argument and return types of the function that will appear later

  • The body-less function declaration is called a functional prototype.

• Example code:

  #include <stdio.h>
  
  // this function prototype is a declaration of sumTo
  int sumTo(int);
  
  int main(void) {
      printf("sumTo(5) is: %d\n", sumTo(5));
      return 0;
  }
  
  // sum integers from 1 to max
  int sumTo(int max) {
      int i, sum = 0;
      for (i=1; i<=max; i++) {
          sum += i;
      }
      return sum;
  }

Page 50:

UNIX Std*

• There are three predefined streams provided to all UNIX programs

  • Standard input (stdin)