Interfaces and process scheduling

Question 1

What are the state of xv6 processes?

Answer 1

Runnable - Process is ready to be executed by CPU.
Running- Process is being executed by CPU.
Sleeping- Process is not runnable, it is waiting for some input/output operation to finish.
Zombie - The process has terminated, but it is waiting for its exit code to be consumed by the parent process.

Question 2

What is the “Round Robin Algorithim”?

Answer 2

Its a process scheduling algorithim where all processes are arranged in a loop, when a “CPU” is looking for the next process to run - it picks the next runnable process in the loop. It runs chosen process until timer interupt, time slice called the quantum usually around 20 ms.

Question 3

what is “process sleep” ?

Answer 3

sometime input/output operation cannot be completed immediatly, i.e waiting for keyboard input, reading data from empty pipe (waiting for the pipe to get data) or even writing into a full pipe (needs to be freed first).
- in such situations the process is temporarly excluded from the scheduling loop(put to sleep)until required condition is fulfilled.

Question 4

What does the process do before going to sleep?

Answer 4

Before going to sleep, the process releases (yeilds) the CPU to the next available
runnable process.

Question 5

What are “blocking System calls”

Answer 5

System calls that do not return until the requested operation is completed and can put the calling process to sleep are called blocking system calls. They block execution of the program until the requested operation completes (or fails).

Question 6

Advantages of round robin algorithim ?

Answer 6

-Simple implementation
-All ready to run processes will eventually be executed.
-Fairness

Question 7

Disadvantages of round robin algorithim?

Answer 7

Poor-performance for Long-Running Process
High context switching overhead
Short Time Slices:

Question 8

what happens in “_entry: In kernel/entry.S” ?

Answer 8

Initialises stack pointer (sp) and calls start(). Each CPU hart gets its own program stack area.

Question 9

What happens in “start() In kernel/start.c” ?

Answer 9

• Enables interrupts;
• Delegates interrupt and exception processing to supervisor mode
  -Gives full memory access to supervisor mode;
• Configures timer interrupts;
• “jumps” to main() using mret instruction, which also switches the CPU hart to supervisor mode

Question 10

What happens in “main() In kernel/main.c” ?

Answer 10

• CPU hart #0 performs most system-wide initialisation work.
• System-wide initialisation of drivers and OS subsystems.
• All other CPU harts wait until the system-wide initialisation is finished and perform only hart-specific initialisations.
• Eventually, all CPU harts
  enter scheduler loop

Question 11

What happens in “kinit() in kalloc.c” ?

Answer 11

Builds system-wide linked list of free memory pages

Question 12

What is a device Driver?

Answer 12

A piece of software that allows the operating system to interact with a particular piece of hardware it is written for.

Question 13

What is Dynamic Driver Loading?

Answer 13

Modern Linux-based operating systems have sophisticated mechanisms that allow them to load or unload driver code when new piece of hardware is connected to the computer

Question 14

Where DL linux drivers stored?

Answer 14

Dynamically loadable Linux drivers are stored in .ko files (Kernel Object files).

Question 15

Shell commands associated with Dynamically loaded linux drivers?

Answer 15

Shell commands lsmod, insmod, rmmod, and modprobe can be used to inspect drivers currently loaded (lsmod) and to remove (rmmod) or add (insmod / modprobe) drivers from Linux kernel “on the fly”.

Question 16

What is RTOS?

Answer 16

Real-Time Operating Systems (RTOS) are tailored for applications with strict time constraints, such as industrial control systems and spaceflight control systems. Their key requirement is predictability in task completion time, with jitter representing variability. In hard real-time OS, any delay in task completion is unacceptable, while in soft real-time OS, occasional delays are acceptable as long as overall deadlines are met. RTOS prioritizes real-time performance guarantees over computational throughput

Question 17

Server OS?

Answer 17

Server OS

Back:

For network servers
Goals: Maximize throughput, requests served, resilience, scalability
Not prioritizing: Startup time, interactive response time
Computational throughput not priority, except in high-performance computing
Recent focus: Power dissipation, energy efficiency

Question 18

Desktop OS?

Answer 18

Typically optimised for rapid startup and rapid interactive response times.
For battery powered devices, energy efficiency is also essential, so avoiding unnecessary processing is a priority

Question 19

General-purpose operating systems?

Answer 19

General purpose OSes, such as Linux, Windows, and
macOS, have to provide certain level of configurability to adapt to the needs of specific application area, e.g.
– Server vs Desktop versions of Microsoft Windows
– Different scheduling algorithms in Linux

Question 20

What is preempptive scheduling?

Answer 20

Making interactive tasks higher-priority ensures their
responsiveness, while making housekeeping tasks lower priority ensures that they do not detract from business performance

Question 21

What is starvation?

Answer 21

The main problem of the “naive” preemptive scheduling with priority queues is that if the high-priority processes are always runnable, lower priority processes never get a chance to run
● This is called starvation and practical scheduling algorithms make efforts to avoid it

One scheme that avoids this problem is Multilevel Feedback Queue Scheduling (MFQS)

Question 22

What are the rules associated with MFQS?

Answer 22

Rule 1: If Priority(A) > Priority(B), A runs (B doesn’t).

Rule 2: If Priority(A) = Priority(B), A & B run in round-robin fashion using the time slice (quantum length) of the given queue.

Rule 3: When a job enters the system, it is placed at the highest priority(the topmost queue).

Rule 4: Once a job uses up its time allotment at a given level (re-gardless of how many times it has given up the CPU), its priority is reduced (i.e., it moves down one queue).

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

Question 23

MFQS attempts to distinguish between interactive and CPU-intensive processes using a heuristic: ????

Answer 23

if the process has consumed a lot of CPU time already, it is probably a CPU-intensive phase and its priority should be lowered.

Question 24

If a process enters interactive phase, its scheduling should be adapted accordingly right? hows does MFQS do this?

Answer 24

MFQS ensures that by periodically resetting
priorities of all processes back to the higest level

Question 25

What is Linux Completly fair Scheduling?

Answer 25

An example of modern scheduling algorithm

Question 26

hat is Virtual Runtime of a process?

Answer 26

Virtual runtime of a process is the total amount of CPU time it has received

“Completely Fair Scheduling” rule: when choosing the next process to run on a CPU, always choose the runnable process with the minimal virtual runtime.

Question 27

Weight Factor?

Answer 27

weighting factor = weight of process with nice value 0 / weight of current task;

where ‘weight’ is roughly equivalent to 1024 * (1.25)^(-nice)