13 IO Systems

1
13 I/O Systems
2
Contents

• I/O Hardware
• Application I/O Interface
• Kernel I/O Subsystem
• Transforming I/O Requests toHardware Operations
• STREAMS
• Performance

3
13.1 Overview

• main jobs of a computer I/O and processing
• I/O subsystem separates the rest of the kernel
  from the complexity of managing I/O devices
• an increasing standardization of software and
  hardware interfaces
• an increasingly broad variety of I/O devices
• device drivers present a uniform device access
  interface to the I/O subsystem

4
13.2 I/O Hardware

• incredible variety of I/O devices
• character versus block
• sequential access, random access
• synchronous, asynchronous
• dedicated, shared
• read-only, read-write
• very slow to very fast

5
Functional categories

• storage devices
• disks, tapes, CD-ROMs
• transmission devices
• network cards, modems
• human-interface devices
• screen, keyboard, mouse

6
Common Concepts

• port a device communicates with the machine via
  a connection point, e.g. a serial port
• bus a set of wires and a protocol
• daisy chain (e.g. SCSI)
• shared direct access (e.g. PCI bus)
• controller a collection of electronics that
  operate a port, a bus, or a device
• a serial-port controller
• a SCSI bus controller (also called host adapter)
• built-in controller (e.g. disk controller)

7
A Typical PC Bus Structure
8
interface registers

• the processor communicates with the controller by
  reading and writing bit patterns in data and
  control registers
• an I/O port typicaly consists of four kind of
  registers
• status register
• control register
• data-in register(s)
• data-out register(s)

9
I/O instruction memory-mapped I/O

• I/O instructions control devices
• controllers registers have addresses, used by
• direct I/O instructions
• memory-mapped I/O
• some system use both techniques
• e.g. the graphic controller in a PC system, I/O
  registers content memory

10
Device I/O Port Locations on PCs
11
13.2.1 Polling

• determines state of device
• command-ready
• busy
• error
• busy-wait cycle to wait for I/O from device
• low efficiency in processor usage

12
13.2.2 Interrupts

• CPU interrupt request line triggered by I/O
  device
• two lines nonmaskable and maskable
• interrupt handler receives interrupts
• interrupt mask to ignore or delay some interrupts
• some unmaskable
• interrupt vector to dispatch interrupt to correct
  handler
• based on priority
• interrupt mechanism also used for exceptions

13
Interrupt-Driven I/O Cycle
14
Intel Pentium ProcessorEvent-Vector Table
15
more about interrupt

• software interrupt, system call, trap
• two part of handler (or a pair of handlers)
• top-half and bottom-half in Linux
• p.61-71 in lab book
• e.g. a series of block read from disk, and copy
  from kernel space to user space

16
13.2.3 Direct Memory Access

• used to avoid programmed I/O for large data
  movement
• e.g. a disk drive does a large data transfer
• requires DMA controller
• a special-purpose processor
• bypasses CPU to transfer data directly between
  I/O device and memory
• handshake between the main processor and the DMA
  controller two wires
• DMA-request wire
• DMA-acknowledge wire

17
Six Step Process to PerformDMA Transfer
18
section review

• a bus
• a controller
• an I/O port and its registers
• handshake between host and controller
• execution of handshake via polling or interrupts
• offload work to a DMA controller for large data
  transfer

19
13.3 Application I/O Interface

• I/O system calls encapsulate device behaviors in
  generic classes
• device-driver layer hides differences among I/O
  controllers from kernel
• devices vary in many dimensions
• character-stream or block
• sequential or random-access
• sharable or dedicated
• speed of operation
• read-write, read only, or write only

20
A Kernel I/O Structure
21
Characteristics of I/O Devices
22
13.3.1 Block and Character Devices

• block devices include disk drives
• commands include read, write, seek
• raw I/O or file-system access
• Memory-mapped file access possible
• character devices include keyboards, mice, serial
  ports
• commands include get, put
• libraries layered on top allow line editing

23
13.3.2 Network Devices

• varying enough from block and character to have
  own interface
• Unix and Windows NT/95/98/2000 include socket
  interface
• separates network protocol from network operation
• includes select functionality
• approaches vary widely
• half-duplex pipes, full-duplex FIFOs, full-duplex
  STREAMS, message queues, mailboxes

24
13.3.3 Clocks and Timers

• provide current time, elapsed time, timer to
  trigger some operation
• programmable interval timer (a hardware) can be
  set to generate periodic interrupts
• process scheduling
• disk synchronization (flush dirty buffers)
• ioctl (on UNIX) covers odd aspects of I/O such as
  clocks and timers

25
13.3.4 Blocking and Nonblocking I/O

• blocking - process suspended until I/O completed
• easy to use and understand
• insufficient for some needs
• nonblocking - I/O call returns as much as
  available
• user interface, data copy (buffered I/O)
• implemented via multi-threading
• returns quickly with count of bytes read or
  written
• asynchronous - process runs while I/O executes
• difficult to program and use
• I/O subsystem signals process when I/O completed

26
13.4 Kernel I/O Subsystem

• scheduling
• buffering
• spooling
• device reservation
• error handling

27
13.4.1 Scheduling

• scheduling a set of I/O requests means to
  determine a good order in which to execute them
• the order in which applications issue system
  calls is usually not optimal
• scheduling improve fairness, reduce the average
  waiting time for I/O
• example disk arm move
• a queue of request for each device
• optimization for overall system performance
  (throughput)
• some OSs try fairness

28
13.4.2 Buffering

• store data in memory while transferring between
  devices or between a device and an application
• to cope with device speed mismatch
• to cope with device transfer size mismatch
• fragmentation and reassembly (in network)
• to maintain copy semantics
• an additional kernel buffer to safeguard
  occasional data change in user space

29
Device Speed Mismatch
Sun Enterprise 6000 Device-Transfer Rates
30
14.4.3 Caching

• caching - fast memory holding copy of data
• always hold just a copy
• key to performance
• buffer versus cache
• buffer may hold the only copy of the data
• cache holds a copy on faster storage of an item
  that resides elsewhere

31
14.4.4 Spooling and Device Reservation

• spooling - hold output for a device
• if device can serve only one request at a time
• i.e., printer
• device reservation - provides explicit facilities
  for exclusive access to a device
• system calls for allocation and deallocation
• watch out for deadlock

32
14.3.5 Error Handling

• OS can recover from transient devices and I/O
  transfer errors
• device unavailable, transient read or write
  failures, network overload
• most return an error number or error code when
  I/O request fails
• system error logs hold problem reports

33
13.4.6 Kernel Data Structures

• Kernel keeps state info for I/O components,
  including open file tables, network connections,
  character device state
• Many, many complex data structures to track
  buffers, memory allocation, dirty blocks
• Some use object-oriented methods and message
  passing to implement I/O
• e.g. Windows NT

34
UNIX I/O Kernel Structure
35
I/O subsystem summary

• management of name of files and devices
• access control to files and devices
• file system space allocation
• device allocation (including reservation)
• buffering, caching, spooling
• I/O scheduling
• device status monitoring, error handling
• device driver configuration

36
13.5 I/O Requests to Hardware Operations

• consider reading a file from disk for a process
• determine device holding file
• translate name to device representation
• physically read data from disk into buffer
• make data available to requesting process
• return control to process
• 10 steps in detail on p. 479-481

37
Life Cycle of an I/O Request
38
13.6 STREAMS

• STREAM a full-duplex communication channel
  between a user-level process and a device
• a STREAM consists of
• - stream head interfaces with the user process
• - driver end interfaces with the device- zero
  or more stream modules between them.
• each module contains a read queue and a write
  queue
• message passing is used to communicate between
  queues

39
The STREAMS Structure
40
13.7 Performance

• I/O is a major factor in system performance
• demands CPU to execute device driver, kernel I/O
  code
• context switches due to block and unblock
• interrupt handling
• data copying
• network traffic especially stressful

41
Intercomputer Communications
example telnet
42
Improving Performance

• reduce number of context switches
• reduce data copying
• reduce interrupts by using large transfers, smart
  controllers, polling (PIO)
• use DMA
• use hardware implementation
• balance CPU, memory, bus, and I/O performance for
  highest throughput

43
Device-Functionality Progression
44
13.8 Summary

• hardware
• buses
• device controllers
• devices
• programmed I/O, and DMA
• system calls
• blocking, or synchronous
• nonblocking, or asynchronous

45
Summary

• kernels I/O subsystem
• I/O scheduling, buffering, spooling, error
  handling, device reservation
• name translation
• I/O system calls are costly
• CPU consumption, context switch, interrupt
  handling, data copy

46
Homework

• paper
• 2, 3, 4
• oral
• 6, 7