Study Question Answers

Chapter #01 -- Introduction


• Question (#1) on Chapter #1: Write 500 words or more describing the interplay between operating system software and the interrupt mechanism. In your writing show that you understand A) how interrupts "drive" the actions of the operating system, and B) how the interrupt mechanism in a modern computing system actually works (what hardware devices do to cause an interrupt and the sequence of events that take place between the time that the currently running program is interrupted and the time that program resumes execution).
  
  
• Answer Ideas:
  
  The essay could touch on some of these points:
  
  
  • Operating system code does not execute continually. It executes only when it is needed. After the boot sequence, the only thing that ever causes OS code to execute is a software interrupt (trap) or a hardware interrupt. That is why we say that the OS is interrupt driven.
    
    
  • In a modern operating system, a user process that needs to do I/O makes a system call to request service from the OS. Typically it does this by first placing parameters in locations where the OS will be able to find them, and then executing a special system call instruction that has the effect of interrupting the CPU.
    
    
  • The interrupt causes a series of switches in the CPU to fire off. The effect is to preserve context of the current process and then cause the CPU to jump to operating system code for servicing the I/O request.
    
    
  • Typically the OS suspends the process P that requested the I/O. It does this by inserting P's process control block (PCB) in a wait-queue associated with the device that has to perform the I/O for P.
    
    
  • If the device is not busy, the OS requests an I/O operation, in behalf of one of the processes in the wait-queue (which could be the process P that just made the system call). Typically the OS makes the request by writing values into registers within the controller of the device and then triggering the start of the operation by sending a signal on a control line of the system bus.
    
    
  • After the OS makes an I/O request, the controller acts independently of the OS until the I/O operation has completed.
    
    
  • After putting P in the device queue and (possibly) making an I/O request to the controller, the OS may take care of some system tasks that need doing. Eventually the OS checks to see if the ready queue is empty. If not, it schedules a process to run and dispatches it to the CPU. If no processes are ready, the OS may do more system tasks, or even execute a 'busy loop' temporarily, until a hardware interrupt comes in.
    
    
  • After a device controller completes an I/O operation, it interrupts the CPU. The effect of this kind of (hardware) interrupt is basically the same as that of the system-call interrupt (trap) described above. After preserving some context of the current process, the CPU jumps into OS code for servicing the interrupt.
    
    
  • In response to the interrupt, the OS may need to copy data from a buffer in the controller to a buffer in primary memory owned by the process P that requested the I/O.
    
    
  • At that point P is probably ready to run, in which case the OS moves it from the I/O wait-queue to the ready queue. Once in the ready queue P will eventually be scheduled by the OS to run again in the CPU.
    
    
  • Typically the OS will then check to see if the wait-queue of the device is empty. If not, the OS will select a process from the queue and request the controller to perform its desired I/O operation.
  
  

Chapter #02 -- Operating Systems Structures


• Question #1 on Chapter #2: Using information gleaned from chapters one and two, write 500 words or more describing the important functions and responsibilities of a modern multi-user, time-sharing operating system.
  
  
• Answer Ideas:
  
  The essay could touch on some of these points:
  
  
  • Operating systems make for more efficient usage of the hardware - helping to keep the CPU and I/O processors doing useful work more often - thus helping make the computer more cost effective - making it pay for itself.
    
    
  • An OS makes the system more convenient for the user - for example by taking over error-prone I/O tasks and providing high level commands that the bare hardware can not provide.
    
    
  • The OS provides a collection of services that are needed in common by all or most programs, and thereby saves all the programmers using the system from having to write code to perform those services -- which would be a lot of extra work.
    
    
  • Operating systems must
    • schedule jobs for execution,
      
      
    • load jobs into memory and set them to running,
      
      
    • operate all the I/O devices on behalf of user programs,
      
      
    • perform context switching in a multiprogramming or time-sharing environment,
      
      
    • manage memory allocation/deallocation for the various user processes,
      
      
    • facilitate the communication and synchronization of concurrent processes that interact,
      
      
    • maintain an on-line file system and all the accompanying structure,
      
      
    • handle swapping (medium term scheduling),
      
      
    • do all the copying and memory allocation required by virtual memory systems,
      
      
    • help protect users from each other - for example it must prevent user processes from accessing each other's primary and secondary memory resources,
      
      
    • protect OS resources from user processes, and
      
      
    • protect network server processes from network-based attacks.
  
  
• Question #2 on Chapter #2: Using information from section 2.7 write 250 words or more describing several different approaches to designing operating system structure. Discuss the advantages and disadvantages of each approach.
  
  
• Answer Ideas:
  
  The essay could touch on some of these points:
  
  
• Operating System Structuring
  
  
  • Some OS's are not very modular
    
    
  • Some are built in layers to make them more modular.
    
    
  • Each layer is only allowed to call upon layers beneath it for services.
    
    
  • layered design facilitates testing.
    
    
  • It's not easy to design layers so that a lower layer never has to ask a higher layer for a service.
    
    
  • Microkernel design puts just a few essential services (say process & memory management plus process communication) in the kernel and the rest is pushed to "user-level"
    
    
  • A recent design innovation involves utilizing dynamically loadable modules. There is a core kernel module resembling a micro-kernel. The dynamic modules can communicate directly with each other through well-defined interfaces.
    
    
  • With microkernel architecture it becomes fairly easy (if not maximally efficient) to emulate many different operating systems "on top of" the microkernel. The idea is to interpret foreign system calls with microkernel system calls.