1. 1  process  process creation/termination  context  process control block (PCB)  context switch  5-state process model  process scheduling short/medium/long term  Unix process model Lecture 4: Process Management

2. 2  process (also called task, or job) is a program in execution  process program code (or text) sequence of instructions to be executed F note: multiple processes may be running the same code (editor, web-browser, etc) context (execution state) Process prompt% ps -efl UID PID PPID C STIME TTY TIME COMMAND root 0 0 0 Jan 19 ? 10:53 swapper root 1 0 0 Jan 19 ? 0:17 init root 2 0 0 Jan 19 ? 7:39 telnetd … mikhail 1546 1321 0 15:38:45 pts/17 3:31 netscape

3. 3 Unix Process Creation  one process can create another process, perhaps to do some work for it child – the new process F an (almost) identical copy of parent (same code, same data, etc.) parent – the original process F parent can either wait for the child to complete, or continue executing in parallel (concurrently) with the child F parent does not terminate until the child does  In UNIX, a process creates a child process using the system call fork( ) In child process, fork() returns 0 In parent process, fork() returns process id of new child child often uses exec( ) to start another completely different program

4. 4 Example of Unix Process Creation #include int a = 6;/* global (external) variable */ int main(void) { int b;/* local variable */ pid_t pid;/* process id */ b = 88; printf("..before fork\n"); pid = fork(); if (pid == 0) {/* child */ a++; b++; } else/* parent */ wait(pid); printf("..after fork, a = %d, b = %d\n", a, b); exit(0); } prompt% forkprogram..before fork..after fork, a = 7, b = 89..after fork, a = 6, b = 88 example execution

5. 5 Unix Process Tree as one processes creates another, they form a tree

6. 6 Process Termination  after completing work a process may execute exit() system call – asking the OS to delete it parent is notified process’ resources are deallocated by operating system  parent may terminate execution of a child process - abort() system call possible reasons F task assigned to child is no longer required F child exceeded allocated resources if parent exits (some) OS require all children to terminate – cascading termination

7. 7 Context  context – information about running process program code (?) static data (bss) heap (dynamic data) procedure call stack - contains temporary data - subroutine parameters, return addresses, local variables register contents F general purpose registers F program counter (PC) — address of next instruction to be executed F stack pointer (SP) F program status word (PSW) — interrupt status, condition codes, etc. OS resources in use - open files, connections to other programs accounting information  note that for each process there are two contexts for user mode and for kernel mode program memory allocation

8. 8 Process Control Block  process control block (PCB) – data structure maintained by OS to keep track of process’ state  contains process id number user id of the owner process state (running, not-running, etc.) PC, SP, PSW, general purpose registers memory limits (static, dynamic) - base/limit register contents list of open files CPU scheduling information (e.g., priority) I/O states, I/O in progress - list of I/O devices allocated to process, list of open files, etc. status of I/O requests …

9. 9 Context Switch  context switch - stopping one process and restarting another sequence of actions  OS takes control (through interrupt)  saves old context in the process PCB  reloads new context from the new process PCB this amounts to changing the execution of processes (as soon as the PC is reloaded)  returns control to app. program  how is context switch different from mode switch? How are they similar? how many mode switches are there in a context switch?  context switch requires several thousand instructions time-sharing OS may do 100–1000 context switches a second

10. 10 Interleaving of Process Execution in Time-Sharing OS process A B C D time process A process B process C process A process B process C process A process C process A process D process C process D process C time from the user’s standpoint multiple processes are executed concurrently in reality OS interleaves process execution

11. 11 Five State Process Model  OS maintains the state of a process.  process is new – when it is being created running - executing on the CPU waiting (blocked) – waiting on input/output even to occur ready – ready to run on the CPU terminated – finished execution, is being destroyed  how many processes in the system can be in a particular state?

12. 12 OS Process Queues OS queues PCBs according to their state  ready queue – processes in state ready (waiting to be run on CPU)  for each device there is an I/O queue – processes in state waiting, blocked on particular device waiting for their request to be submitted to the device controller

13. 13 Short Term Scheduler  selects process from the ready queue to run on the CPU  runs when process is created or terminated process switches from running to blocked interrupt occurs  goals: minimize response time (e.g., program execution, character to screen) minimize variance of average response time — predictability may be important maximize throughput F minimize overhead (OS overhead, context switching, etc.) F efficient use of resources fairness — share CPU in an equitable fashion

14. 14 Other Types of Schedulers  medium-term scheduler temporarily swaps processes in/out of main memory may suspend/resume processes goal: balance load for better throughput  long-term scheduler (job scheduler) – does not exist on time-sharing OS selects job from the spool, and loads it into memory executes infrequently, maybe only when process leaves system controls degree of multiprogramming goal: good mix of CPU-bound and I/O-bound processes F I/O-bound process – spends more time doing I/O than computations, many short CPU bursts F CPU-bound process – spends more time doing computations; few very long CPU bursts

15. 15  enhanced 5-state model  blocked and ready states has to be split depending on whether a process is swapped out on disk or in memory  running state is also split depending on the mode: kernel or user  Unix process states: created - just created not yet ready to run ready (memory) - ready as soon as kernel schedules it ready (disk) - ready, but needs to be swapped to memory asleep - blocked (memory) - waiting on event in memory asleep - blocked (disk) - waiting on event on disk running (kernel) - executing in kernel mode running (user) - executing in user mode zombie - process terminated but left a record for parent to collect Unix Process States

16. 16 Unix Process Scheduling  process is running in user mode until an interrupt occurs or it executes a system call  if time slice expires the process is preempted and another is scheduled  a process goes to sleep if it needs to wait for some event to occur and is woken up when this event occurs  when process is created decision is made whether to put it in memory or disk zombie ready kernel running ready asleep created user running user mode kernel mode disk wakeup swap out wakeup swap in swap out not enough memory enough memory fork exit sleep schedule return interrupt, system call interrupt preempt

17. 17 Lecture Review  the execution of a certain task from OS standpoint is a process OS maintains (creates/destroys) processes for user OS schedules processes for user F there are three types of schedulers: short/medium/long term  OS saves process states in a process control block  OS context switches between processes  in 5-state process model a process can be in one of five states: created, running, ready, blocked, terminated UNIX adds states for being swapped out to disk depending on process state its PCB is attached to either ready queue or device queue