1. spcl.inf.ethz.ch @spcl_eth S ALVATORE DI G IROLAMO (TA) Networks and Operating Systems: Exercise Session 1

2. spcl.inf.ethz.ch @spcl_eth About me:  Name: Salvatore Di Girolamo  Contact: digirols@inf.ethz.chdigirols@inf.ethz.ch  Office: CAB E66 What will we do?  Remark important concepts + exercises  It is a cooperative session! Assignments:  Not mandatory, not graded  You get a feedback if you hand-in them (see Git tutorial) 2 Introduction

3. spcl.inf.ethz.ch @spcl_eth What is an Operating System? 3 Glue Referee Illusionist Which are the main roles of an OE? It’s an abstraction layer between the user and the hardware resources Referee: sharing, protection and communication Illusionist: create the illusion of a real resource (virtualization) Glue: Provide high-level abstraction, hiding details of hardware

4. spcl.inf.ethz.ch @spcl_eth General OS structure Kernel Privileged mode User mode Application System Library Application System Library Server process (daemon) Server process (daemon) System Library System calls CPU Device CPU 4

5. spcl.inf.ethz.ch @spcl_eth OS Kernel  That part of the OS which runs in privileged mode  Large part of Unix and Windows (except libraries)  Small part of L4, Barrelfish, etc. (microkernels)  Does not exist in some embedded systems  Kernel is just a (special) computer program.  Typically an event-driven server.  Responds to multiple entry points:  System calls  Hardware interrupts  Program traps  May also include internal threads. 5

6. spcl.inf.ethz.ch @spcl_eth OS Kernel 6 Embedding applications in the kernel? Pros: PerformanceCons: Abstraction, Security, Bloated OS How kernel/user space distinction ensures security? User-space applications cannot issue all type of instructions Hardware devices can be accessed only in kernel mode Interrupts can be enabled/disabled only in kernel mode Btw, what happens if they could be managed in user mode? Which of the following operations should be privileged? Set a timer Turn off interrupts Create a process Issue a trap Read the clock

7. spcl.inf.ethz.ch @spcl_eth Recall: System Calls  RPC to the kernel  Kernel is a series of syscall event handlers  Mechanism is hardware-dependent System calls Privileged mode User mode User process runs Process resumes Execute kernel code Execute syscall 7

8. spcl.inf.ethz.ch @spcl_eth System calls 8 What is the purpose of system calls? -Request services to the OS How would you create a new process? NAME fork - create a child process SYNOPSIS #include pid_t fork(void); DESCRIPTION fork() creates a new process by duplicating the calling process. The new process, referred to as the child, is an exact duplicate of the calling process, referred to as the parent … NAME fork - create a child process SYNOPSIS #include pid_t fork(void); DESCRIPTION fork() creates a new process by duplicating the calling process. The new process, referred to as the child, is an exact duplicate of the calling process, referred to as the parent … -Enable the OS abstraction -e.g., memory virtualization NAME execl, execlp, execle, execv, execvp, execvpe - execute a file SYNOPSIS #include int execv(const char *path, char *const argv[]); DESCRIPTION The exec() family of functions replaces the current process image with a new process image. The functions described in this manual page are front-ends for execve(2).

9. spcl.inf.ethz.ch @spcl_eth Process ingredients  Virtual processor  Address space  Registers  Instruction Pointer / Program Counter  Program text (object code)  Program data (static, heap, stack)  OS “stuff”:  Open files, sockets, CPU share,  Security rights, etc. 9

10. spcl.inf.ethz.ch @spcl_eth Process address space Stack Text Data BSS 00000000 7FFFFFFF (addresses are examples: some machines used the top address bit to indicate kernel mode) Should look familiar … 10

11. spcl.inf.ethz.ch @spcl_eth Process lifecycle runnable (ready) running blocked (waiting) terminated created dispatch preemption I/O operation I/O completes runnable (ready) running runnable (ready) blocked (waiting) blocked (waiting) running runnable (ready) runnable (ready) 11

12. spcl.inf.ethz.ch @spcl_eth  A process may create another process  Parent-Child relation  Process tree  Unix: fork()/exec()  The fork returns different values:  At the parent: the Process ID (PID) of the children  At the children: zero  Both parent and child execute the same next line!  The variables until the fork() are shared  Usually the child executed an exec() to execute a different code  Parent has to wait for the completion of its children  wait/waitpid  Otherwise you create zombies!!! 12 Processes Creation What is the return value of this function?