Operating Systems Lecture 4 Computer Systems Review Read: Chapter 2

1
Operating SystemsLecture 4Computer Systems
ReviewRead Chapter 2
2
Computer System Components
Computer system components CPU I/O controllers
(e.g. disk controller) Memory Controller The
CPU and device controllers can operate
concurrently. The Memory controller ensures
orderly access to shared memory.
3
Computer-System Architecture
4
Interrupt Driven O.S.
Most modern operating systems are interrupt
driven. Start-up 1) Load O.S. (kernel) and start
it running. 2) O.S. waits for an
event (an interrupt). Definition of
interrupt An interrupt is a method to ensure
that the CPU takes note of an event. Types of
interrupt Hardware interrupts (e.g. from an I/O
device) Software interrupt (from an executing
process. System Call) Also called a trap or an
exception Generated to signal error (e.g. divide
by zero) or to request service from OS (e.g.
I/O).
5
Interrupt Handling

• The operating system preserves the state of the
  CPU by storing registers and the program counter.
• The O.S. determines which type of interrupt has
  occurred by one of two methods
• polling each device ("Did you interrupt me?")
• vectored interrupt system (Table of addresses)
• Separate segments of code determine what action
  should be taken for each type of interrupt.
• The OS executes the sequence of commands
  associated with the given interrupt.
• The OS recovers the stored information from the
  original process and continues execution.

6
Interrupt Time Line For a Single Process Doing
Output
7
Synchronous I/O

• Synchronous I/O
• After I/O starts, control returns to user
  program only upon I/O completion.
• Process requests I/O
• CPU sends I/O request to device controller
• CPU waits for I/O to finish (signalled by
  interrupt)
• Continue executing original process (after
  handling interrupt)
• At most one I/O request is outstanding at a time,
  no simultaneous I/O processing.
• Problem I/O is slow. CPU is idle while waiting
  for I/O to finish.

8
Asynchronous I/O

• Asynchronous I/O
• After I/O starts, control returns to user program
  without waiting for I/O completion.
• Process requests I/O
• CPU sends request to device controller
• CPU continues to process instructions (either
  previous process or another one).
• Interrupt signals I/O completion
• CPU processes interrupt
• Less time is used to process I/O requests using
  asynchronous I/O than with synchronous I/O.

9
Two I/O Methods
Synchronous
Asynchronous
10
Other Considerations
When using asynchronous I/O, there may be
multiple I/O requests at the same time. Requests
for multiple devices Multiple requests for a
single device. A device-status table keeps track
of the requests. It stores Device type Device
address Device State (not-functioning, idle,
busy) Wait queues list multiple requests for a
single device.
11
Device-Status Table
12
Direct Memory Access Structure

• If only 1 character is transferred per I/O
  request, then an interrupt is generated after
  each character is transferred.
• Slow I/O device (e.g. modem)
• 1 ms/char 1000 ms/char
• Time needed to process interrupt 2 ms
• This leaves 998 ms for processing other things.
• Fast I/O device (e.g. hard disk)
• 4 ms/char
• Only leaves 2 ms to process other things.
• Solution Direct Memory Access (DMA)
• Used for high speed I/O devices.
• Device controller transfers blocks of data from
  buffer storage directly to main memory without
  CPU intervention
• Only one interrupt is generated per block, rather
  than the one interrupt per byte.

13
Storage Structure
The Fetch-Execute Cycle Computer programs must
be in main memory to be executed. The CPU fetches
the next instruction (from the address stored in
the instruction counter) The CPU executes the
instruction This may involve fetching operands
from main memory. It may also involve storing
the result in main memory. Ideally, we would
want all programs stored in main memory. This is
impractical because Main memory is too small to
store all needed programs and data. Main memory
is volatile. It loses its contents when the
power goes off.
14
Secondary Storage

• Main memory only large storage media that the
  CPU can access directly.
• Secondary storage extension of main memory that
  provides large nonvolatile storage capacity.
• Most common type of secondary storage
• Magnetic disks rigid metal or glass platters
  covered with magnetic recording material
• Disk surface is logically divided into tracks,
  which are subdivided into sectors.
• The disk controller determines the logical
  interaction between the device and the computer.
• Other types of storage Cache, CD_ROM, Magnetic
  tape.

15
Moving-Head Disk Mechanism
16
Storage Hierarchy

• Storage systems organized in hierarchy.
• Speed
• Cost
• Volatility
• The fastest memory is the most expensive and most
  volatile. Used for CPU registers.
• As move down the hierarchy, memory gets slower
  and cheaper. Non-volatile memory is the slowest.

17
Storage-Device Hierarchy
18
Caching

• Accessing main memory can be slow compared to the
  time it takes to execute instructions.
• Caching is the use of high-speed memory to hold
  recently-accessed data.
• Blocks of instructions or data from main memory
  are transferred to a smaller, faster memory the
  cache.
• Locality of reference ensures efficient use of
  the cache
• As a process executes, it references memory in
  clustered areas.
• Nearby memory locations are likely to be accessed
  in the near future.
• Therefore, when one item is requested from main
  memory, an entire block is transferred to the
  cache.
• Use of a cache requires a cache management
  policy.
• Caching introduces another level in storage
  hierarchy. This requires data that is
  simultaneously stored in more than one level to
  be consistent.

19
Maintaining Consistency
An operating system needs to maintain consistency
at different levels of the storage
system. Example Suppose an integer, A, is
stored in a file, B, on the disk. Suppose we
want to increment A by 1. A is copied from disk
to main memory. A is copied from main memory to
cache A is copied from the cache to a
register. We now have 4 copies of A. A is
incremented in the register. Now A has different
values in different locations. The OS must ensure
that any process wanting to access A gets the
most recently updated version.
20
Migration of A From Disk to Register