Review for the Final Exam

All chapter #s are for the 9^th edition.

Chapter 1

What is an O.S?
The kernel
A traffic cop
A resource allocator
The kernel + utilities and stuff

Chapter 2

• System Calls

  • It’s a request to the OS to do something

  • It’s

    • Load registers

    • TRAP

    • Answer in a return register

• I/O Overlap

  • Why

  • How to achieve

Chapter 3

• What is a process

  • Different from thread

  • Different from executable

• Process queues

  • Running

  • Ready

  • New

  • Exit

  • Wait

• Ways to make a new process

  • Fork

  • Spawn/system

Chapter 4

• Threads -- definition of

• Threads share code, heap(global variables and new/delete stuff) but not stack(local variables)

• C++ Threads

  • http://euclid.nmu.edu/~rappleto/Classes/CS426/Threads/C/NewStyle/new_threads.cc

• Java threads

  • http://euclid.nmu.edu/~rappleto/Classes/CS426/Threads/Java/UpDown/ud3.java

• locking

  • Even readers need to lock if there is anyone writing.

Chapter 5 (Skip)

Chapter 6

• CPU Burst -- definition

• Preemptive scheduling

• Aging

• Criteria

  • CPU Utilization

  • Throughput

  • Wait time

  • Starvation

  • Fairness

• Algorithms

  • FCFS

  • Shortest Job Next

  • Round Robin

  • Priority

  • Multi-level feedback queues (yes, really)

  • Real Time

    • Priority Inversion -- the definition and the solution

    • Which process goes next

      • Most important????

      • Soonest deadline???

    • Soft vs Hard real time

Chapter 7

• Deadlock

  • Definition

  • Avoid by ordering resources

  • Avoid by using two phases

  • Avoid by drawing the wait-for graph

  • Avoid by timeouts

Chapter 8 and 9

• Definition of swapping

• Segments

  • Hardware needed

  • External fragmentation

  • Choosing a spot

    • First fit

    • Best fit

    • Worst fit

• Paging

  • Hardware needed

  • What do do on an MMU interrupt (see chart 9.2.2, page 405)

  • Page Tables

    • Protection

    • Computing the size

    • Multi-level page tables

  • Page table base registers

  • Translation lookaside buffer

  • Internal fragmentation

  • Sharing pages

    • Code

    • Heap (if you’re two threads of the same process)

    • copy-on-write

  • Replacement algorithms

    • Optimal (Belady's Min)

    • LRU

    • FIFO

    • Additional bits

Chapter 10

• Disk scheduling algorithms

  • FCFS

  • Elevator

  • Cyclic elevator

  • Shortest seek next

• RAID (especially levels 0, 1, 5)

  • What to do for error

  • What to do when a new disk is inserted

  • Speed

    • Small reads

    • small writes

    • large reads

    • large writes

  • Capacity

  • Sledge hammer factor

File systems (Chapter 11)

• Files

  • Have data

  • Have metadata - name, permissions, times, locations on disk, size ….

• File Operations

  • Open – finds the file, check permissions, expensive

  • Create – Make a new file

  • Unlink – delete a file or reduce it’s reference count

  • Close – Free resources, maybe cheap

  • Read – get data off disk or from cache

  • Write – often just store in cache and schedule the write to hard storage

  • Append – might be nice to have, but we don’t

  • Seek – move to new location in the file

  • Stat – Get metadata

  • Truncate – cut file back to N bytes. Often back to zero bytes.

• Directory Operations

  • Create a file

  • Delete a file

  • Rename a file

  • opendir/readdir/closedir

• Directory Structure

  • Sometimes just one directory – sucks

  • Sometimes are trees

  • Sometimes are full graphs

• Partitions – divide the hard drive into chunks and treat each chunk a a separate file system

• Mounting is putting a file system on top of a directory

  • UNIX does this

  • Windows sometimes uses drive letters instead.

• Consistency Semantics (11.5.3)

  • Session (open/close) consistency

  • Unix (single image) consistency

  • Immutable consistency

  • Ad hoc consistency (not in the book)

File systems

• VFAT (http://euclid.nmu.edu/~rappleto/Classes/CS426/Notes/FileSystems/vfat-file-system.html)

  • VFAT organization

    • Master boot record

    • Two vfats

    • root directory

    • data stuff

  • File names and long file names

• EXT2 (http://euclid.nmu.edu/~randy/Research/Papers/EXT2/)

  • EXT2 organization

    • Many groups .. each group has a

      • superblock – stuff about the whole file system like free space in file system

      • group descriptor block – stuff about that block like free space in this block

      • inode bitmap – which inodes are free

      • block bitmap – which blocks are free

      • data blocks

    • root directory is always inode #2

    • directories are – list of (file_name_length, file_name, inode_number)

    • inodes are

      • metadata like permissions and times

      • a list of 13 pointers to data

      • a pointer to an indirect block

      • a pointer to a double-indirect block

• How to program a shell

  • no fork bombs please

  • iterate through the path and try to exec

  • dup2 is used for pipe and file redirection

  • Can catch ^C using a signal handler

• Opendir/Readdir/Cloasedir

• Supercomputers

  • Are zillions of cores (and maybe older cores too)

  • Programming them is very hard

  • The network costs as much as the computers

  • Use 5-15 MW of power