Processes
Processes
 A process is a program in execution; process
execution must progress in sequential fashion. Can
be characterized by its trace.
 The OS interleaves the execution of several
processes to maximize processor utilization
 A process requires resources, which are managed
by the operating system
 OS supports Inter-Process Communication (IPC)
and user creation of processes
2
Trace of Process
 Sequence of
instructions
that execute
for a process
 Dispatcher
switches the
processor
from one
process to
another
3
Time slice ~= 1/10th of a second
4
Process state models
 The OS is responsible for interleaving the
execution of the processes.
 It is important to understand the possible
behaviors of all the processes and how to react to
them.
 The behavior of processes is described by means of
the process state model.
5
Two-State Process Model
 Process may be in one of two states
 Running
 Not-running
6
Queuing Diagram
Enter
Queue Dispatch
Processor
Exit
Pause
(a) Queuing diagram
 Dispatcher
 Program that assigns the processor to one process or
another
 Prevents a single process from monopolizing
processor time
7
Reason for Process Creation
8
Reasons for Process Termination
9
Reasons for Process Termination
10
5 Process States
 As a process executes, it changes state
 New: The process is being created.
 Running: Instructions are being executed.
 Waiting or blocked: The process is waiting for some
event to occur.
 Ready: The process is waiting to be assigned to a
process.
 Terminate or Exit: The process has finished execution.
11
Diagram of Process State
12
Example for 3 Processes
13
OS Control Structures
Process Control Block (PCB)
 Information associated with each process.
 Process ID
 Process state
 Program counter (PC)
 CPU registers (Context data)
 CPU scheduling information (Priority)
 Memory-management information (Memory pointer)
 Accounting information (Time, acct. #)
 I/O status information (requests, devices assigned)
15
Process Control Block
 Contains the process attributes
 Created and manage by the operating system
 Allows support for multiple processes
Process = Program code + Data + PCB
16
Suspending Processes
17
Two Suspended States
18
Process Scheduling Queues
 Job queue – set of all processes in the system.
 Ready queue – set of all processes residing in main
memory, ready and waiting to execute.
 Device queues – set of processes waiting for an I/O
device.
 Process migration between the various queues.
19
Ready Queue, I/O Device Queues
20
Process Scheduling
21
Schedulers
 Long-term scheduler: job scheduler
 selects which processes should be brought into the ready
queue.
 invoked infrequently (seconds, minutes)
 controls the degree of multiprogramming
 Medium-term scheduler
 allocates memory for process.
 invoked periodically or as needed.
 Short-term scheduler: CPU scheduler
 selects which process should be executed next and allocates
CPU.
 invoked frequently (ms)
22
Process Switching
 Definition
 The activity that occurs when the OS kernel switches
between processes in an effort to share the CPU among
competing, runable processes
 Actions
 Save contents of hardware registers (PC, SP, ...)
 Load PC with location of resume point
 Load hardware registers with new context if full context
switch
23
Process Switching
 Process-switch time is overhead
 CPU utilization affected by quantum and process-switch
time
 Quantum: Time-slice (max CPU interval) given to a
process
 Time dependent on hardware support
 Types
 Voluntary: Process blocks
 Involuntary: End of time quantum or higher-priority,
runable process gets control of CPU
24
CPU Process Switch
25
Mode switch vs Process switch
 When a mode switch occurs, the PCB of the
process that was running is still marked Running
and is not placed in the Blocked or Ready queue
 However, depending on what caused the mode
switch, a process switch could follow
26
Modes of execution
 User mode:
 Some parts of virtual address space can not be accessed
 Some instructions (e.g., memory management) can not
be executed
 Kernel mode:
 Can access kernel address space
 Is a fixed part of the virtual address space of every process
 System call puts user into kernel mode
27
Mode, Space and Context
Process Context
Kernel Context
System call
Application
System calls
Exceptions
(user space)
(user + system space)
X
not allowed
User Mode
Interrupts,
System tasks
(user space)
Kernel Mode
More privileges
Interrupt occurs
28
Execution of the OS
29
Execution within user processes
30
Unix Process Creation
 Parents create children
 results in a tree of processes
 Resource sharing
 Parent and children share all resources
 Children share subset of parent’s resources
 Parent and child share no resources
 Execution
 Parent and children execute concurrently
 Parent waits until children terminate
31
Unix Process Creation
 Address space
 Child duplicate of parent
 Child has a program loaded into it
 UNIX examples
 fork system call creates new process
 exec family of system calls used after a fork to replace
the process’ memory space with a new program
32
Unix Process Creation (fork)
 Allocate new PID and proc structure
 Init child’s process control block (PCB)
 Copy parent’s registers to child’s hardware context
 Copy parent’s process image to child’s
 Mark child runnable and give to scheduler
 Return PID = 0 to child
 Return child PID to parent
33
Unix Process Creation (fork)
 Effects of fork
 Child gets a copy of its parent’s memory
 BUT changes made by child are not reflected in parent
 EXCEPT a child can affect the I/O state of parent

File descriptors can point to same file table entry
 WARNING: Parent should “fflush(stdout)” before fork if
using printf()
34
exec Operations
 Use to overlay process image with new program
 Find executable file (argv[0])
 Copy exec args and env variables to kernel space
 Free memory currently allocated to process image
 Allocate new memory for process image
 Copy exec args and env variables into new space
 Begin executing in main()
35
exec Functions
 Six exec functions:
 execl
 execv
 execle
 execve
 execlp
 execvp
l  list of arguments
v  argv[] vector
p  file name
e  envp[] vector instead of
inheriting environment of parent
36
waitpid
 pid_t waitpid(pid_t pid, int *status, int options);
 Unlike wait(2), waitpid doesn’t have to block
 wait(2) blocks until one of its children terminates
37
waitpid – pid parameter
 pid == -1: Wait for any child to terminate
 pid > 0: Wait for child whose PID equals pid
 pid == 0: Wait for any child whose PGID equals the
PGID of process pid
 pid < -1: Wait for any child whose PGID = abs(pid)
38
waitpid – status parameter
 Contains exit status, signal number, and flags
 WIFEXITED(status)
 True if child terminated normally
 Ex: printf(“exit = %d\n”, WEXITSTATUS(status));
 WIFSIGNALED(status)
 True if child terminated and didn’t catch signal
 Ex: printf(“signo = %d\n”, WTERMSIG(s));
 WIFSTOPPED(status)
 True if child is STOPPED
 Ex: printf(“signo = %d\n”, WSTOPSIG(s));
39
waitpid – options parameter
 options = 0
 No special handling
 options = WNOHANG
 Don’t block if child is not available and return 0
 Ex:
rc = waitpid(pid, &status, WNOHANG);
if (rc == -1) { . . . error . . . }
else if (rc == 0) { . . . no child available . . .}
else { . . . rc should equal pid . . . }
40
waitpid options
 options = WUNTRACED
 Return status of child if child stopped and status has not
already been returned (assumes job control support)
41