image-20221229232225217

fork() 创建一个和父进程一样的子进程，然后在子进程中return 0

父进程中return子进程的pid

exec()覆写当前进程的代码段，堆栈等，成功的call to exec从不返回

Syscall and Trap

How OS keep control of hardware and resources?

At Boot time, kernel intialize trap table for CPU.

image-20221230000129385

image-20221230002139714

Trap: User mode -> Kernel Mode

  • Save Reg to kernel stack (hardware)
  • move to kernel mode
  • jump to trap handler

Return-from-trap: Kernel mode -> User mode

  • Restore Reg from kernel stack (hardware)
  • move to user mode
  • jump to process next instruction

Context Switch

image-20221230001632930

image-20221230002509743

Save context: Regs, PC, Kernel stack pointers.

Since Interrupt is handled by hardware, which means OS did not have control at that time, trap/return-from-trap handler will need kernel stack to stores those information so when OS take control, it would be able to get regs value.

image-20221230011337763

Scheduling

$T_{turnaround} = T_{complete} - T_{arrival}$

$T_{response} = T_{first run} - T_{arrival}$

FIFO

SJF (Short Job First)

在同时到达的process中选择最短的

STCF (Shortest Time-to-Completion First)

抢占式的SJF

Good for Turnaround time

Bad for Response Time

RR(Round Robin)

Good for Response Time

Bad for Turnaround Time

I/O

image-20230107125008931

MLFQ(Multi-Level Feedback Queue)

  • Rule 1: if $Priority(A) > Priority(B)$, A runs B doesn’t

  • Rule 2: if $Priority(A) = Priority(B)$, A and B run in RR

  • Rule 3: When a job enters the system, it is placed at the highest priority

  • Rule 4a: If a job uses up its time allotment at a given level, its priority is reduced

    Here allotment meas: Each level of queue has unique length of time slice

    image-20230107131834905

  • Rule 5: After some time period S, move all jobs in the system to the topmost queue

image-20230107131854933