Working with the Unix OS

More documents

Recommendations

Info

Process Scheduling 12. PROCESS SCHEDULING The Scheduler The kernel is responsible for sharing CPU time between competing processes. Multi-Level Priority Queue - linked list of runnable processes Processes are allocated CPU time in proportion to their importance. Time is allocated in fixed size units called "time quantums" (~ 1/10 second). Process Table Next PID PPID Stat MLPQ ---> . 36 12 R . ---> Process 34 | -- free entry 0 .-- | - 18 1 S 1 .-- | free entry 2 - | -> - 12 1 R . ---> Process 12 3 - | free entry 4 - | -> - 48 1 S ----> - 1 - R . ---> Process 1 Scheduling Rules - Every second, scheduler recalculates priority of all runnable processes - organizes them into priority queues. - Every 1/10 sec, the scheduler selects highest priority process in priority queue and allocates it the CPU. - If process is runnable at end of time quantum, it is placed at end of its priority queue. - If process sleeps on an event during time quantum, the scheduler selects next runnable process. - If process returns from system call during time quantum, and higher priority process is ready to run, the lower priority process is preempted. - Every hardware clock interrupt (1/100 second), the process's clock tick count is incremented, every 4th tick, scheduler recalculates priority. priority = K1 / (recent CPU usage) + K2 / (nice setting) - A process's priority diminishes if it uses a lot of CPU in a window of time. - An interactive process waits for a user to press a key, it uses no CPU time and thus its priority level rises. Thus interactive processes obtain good response times. Memory Management Sharing of RAM between processes (secure, efficient) ! Memory Pages Allow processes bigger than RAM capacity to execute. RAM (code, data, stack) divided into fixed-size pages, analogous to, disk divided into fixed-size blocks. The size of memory page is set to size of disk block. Only pages of process, currently accessed or recently accessed are stored in RAM pages, the rest are on disk. 169
Process Scheduling Page Tables and Regions Process Table --- --- --- User area Region table to code page --------------- to data page --------------- to stack page Process Info current dir unmask value pending signal control term Code page table ------------ ------------ -- Data page table ------------ ------------ -- Stack page table ------------ ------------ -- RAM/disk pages of code RAM/disk pages of data RAM/disk pages of stack Process Scheduling The kernel allocates time slices/quantum, preempts the process & schedules another when time slice expires, then reschedules. Clock time 50/100 times a second - interrupt UNIX uses "round robin with multilevel feedback", process through many iteration of feedback loop. algorithm schedule_process input: none output: none { while (no process picked to execute) { for (every process on run queue) pick highest priority process that is loaded in memory; if (no process eligible to execute) idle the machine; /* interrupt takes machine out of idle state */ } remove chosen process from run queue; switch context to that of chosen process, resume its execution; } Figure 46. Process Scheduling It makes no sense to select a process if it is not loaded in memory, cannot execute until swapped in. If several processes tie for highest priority, pick the one that has been "ready to run" the longest. Each process table entry has a priority field. The priority of a process in user mode is a function of its recent CPU usage (recently used lower priority). User & Kernel mode priority - The kernel does not change the priority of processes in kernel mode. 170
Page 1 and 2:
UNIVERSITY OF ATHENS DEPARTMENT OF
Page 3 and 4:
Figure 27. Driver Entry Points.....
Page 5 and 6:
Introduction to Unix ! Execution ma
Page 7 and 8:
Introduction to Unix It works in th
Page 9 and 10:
Introduction to Unix * matches 0 or
Page 11 and 12:
Introduction to Unix vi file-name s
Page 13 and 14:
Introduction to Unix 1,$s/^/>>/ 1,$
Page 15 and 16:
Introduction to Unix ! Special Devi
Page 17 and 18:
Introduction to Unix Working with F
Page 19 and 20:
Introduction to Unix ! Communicatio
Page 21 and 22:
Unix Shells 2. UNIX SHELLS Back Quo
Page 23 and 24:
Unix Shells ! CASE Statement case t
Page 25 and 26:
Unix Shells two lines !Y@Y@Z@Z@ZY c
Page 27 and 28:
Unix Shells if $append then tee -a
Page 29 and 30:
C Programming - Basics main() { int
Page 31 and 32:
C Programming - Basics e.g. i += 2
Page 33 and 34:
External Variables A C program cons
Page 35 and 36:
C Programming - Basics main.c #incl
Page 37 and 38:
C Programming - Basics default: pri
Page 39 and 40:
C Programming - Basics struct tnode
Page 41 and 42:
Unix Tools 4. UNIX TOOLS Tools of t
Page 43 and 44:
Unix Tools Korn shell ! .profile HI
Page 45 and 46:
Unix Tools grep "$name" $PHONEBOOK
Page 47 and 48:
Unix Tools awk '$3 > 0 { print $1,
Page 49 and 50:
Unix Tools END { print max, maxline
Page 51 and 52:
Unix Tools Let an example input fil
Page 53 and 54:
Unix Tools Brown eqn 1 # output Jon
Page 55 and 56:
Development Tools Make looks at the
Page 57 and 58:
Development Tools /* compute size o
Page 59 and 60:
Development Tools #endif #line cons
Page 61 and 62:
Development Tools 1.1 5 lines Retri
Page 63 and 64:
Development Tools In Emacs, you may
Page 65 and 66:
Development Tools */ %{ #define YYS
Page 67 and 68:
C Libraries 6. C LIBRARIES Input an
Page 69 and 70:
C Libraries } while (!feof(fp1)) pu
Page 71 and 72:
C Libraries } free(p); /* what is w
Page 73 and 74:
C Libraries } if ((fp->flag&( _READ
Page 75 and 76:
C Libraries /* dirwalk: apply fcn t
Page 77 and 78:
C Libraries } switch (pid=fork()){
Page 79 and 80:
C Libraries What is curses? - libra
Page 81 and 82:
C Libraries } if ((fd=fopen(argv[1]
Page 83 and 84:
Introduction to kernel 7. INTRODUCT
Page 85 and 86:
82 Introduction to kernel Programs
Page 87 and 88:
Introduction to kernel } { printf("
Page 89 and 90:
Introduction to kernel The kernel r
Page 91 and 92:
Introduction to kernel UNIX Interna
Page 93 and 94:
Introduction to kernel File System
Page 95 and 96:
Introduction to kernel Every proces
Page 97 and 98:
Processes I Process States 1. Execu
Page 99 and 100:
Processes I } printf("%d %s\n", cnt
Page 101 and 102:
Processes I dup (and other similar
Page 103 and 104:
Processes I The u area contains - p
Page 105 and 106:
Processes I The Context of a Proces
Page 107 and 108:
Processes I Process Control use and
Page 109 and 110:
Processes I Although the processes
Page 111 and 112:
Processes I Handling Signals - proc
Page 113 and 114:
Processes I USER=neville PATH=/user
Page 115 and 116:
Processes I ! Shadow Password /etc/
Page 117 and 118:
struct timeval { long tv_sec; /* se
Page 119 and 120:
Processes I F_SETFL set status flag
Page 121 and 122: Processes I process is the child pr
Page 123 and 124: Processes I Rules about sharing of
Page 125 and 126: Processes II 9. PROCESSES (II) FORK
Page 127 and 128: Processes II /* background.c */ #in
Page 129 and 130: Processes II } while (1){ printf("p
Page 131 and 132: Processes II } /* child */ signal (
Page 133 and 134: Processes II int readline(int fd, c
Page 135 and 136: Processes II { } char process_name[
Page 137 and 138: Processes II - Pipe in a single pro
Page 139 and 140: Processes II System V IPC - message
Page 141 and 142: Processes II int msgctl(int msqid,
Page 143 and 144: Processes II For every semop operat
Page 145 and 146: Processes II /* client loop */ mesg
Page 147 and 148: I/O Subsystem Drivers frequently sl
Page 149 and 150: I/O Subsystem Terminal Drivers - In
Page 151 and 152: I/O Subsystem Canonical Mode algori
Page 153 and 154: I/O Subsystem Raw Mode Raw mode is
Page 155 and 156: Figure 43. Pushing a Module onto a
Page 157 and 158: Interprocess Communication } { ptra
Page 159 and 160: Interprocess Communication T ID KEY
Page 161 and 162: Interprocess Communication }; int m
Page 163 and 164: Interprocess Communication for (i=0
Page 165 and 166: Interprocess Communication /* struc
Page 167 and 168: Interprocess Communication ! Compar
Page 169 and 170: … connect(sockfd,(struct sockaddr
Page 171: Interprocess Communication Address
Page 176 and 177: Figure 49. Tie breaker rule Process
Page 178 and 179: Process Scheduling Clock - restart
Page 180 and 181: Process Scheduling Figure 57. Mappi
Page 182 and 183: Figure 62. Sequence of Swapping Ope
Page 184 and 185: Buffer Cache The kernel takes buffe
Page 186 and 187: Buffer Cache The kernel places the
Page 188 and 189: Buffer Cache Process B will find th
Page 190 and 191: Unix Administration 14. UNIX ADMINI
Page 192 and 193: Unix Administration Example output
Page 194 and 195: Unix Administration *) IN=-mtime -2
Page 196 and 197: Unix Administration Load /Unix kern
Page 198 and 199: Unix Administration for (i=0; i&i d
Page 200 and 201: Unix Administration fi done echo $s
Page 202 and 203: Unix Administration New Software !
Page 204 and 205: Unix Security Example: group name:g
Page 206 and 207: Unix Security $ crypt < exam320.Z >
Page 208 and 209: Unix Security - exported filesystem
Page 210 and 211: Unix Security system(cmdstr); $ ech
Page 212: Unix Security - repairs damaged use
show all

Working with the Unix OS

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?