I/O Systems

Maciej Sałata 2006

I/O Systems

PART I – Abstract aroach I/O Hardware Application I/O Interface Kernel I/O Subsystem Transforming I/O Requests to Hardware

Operations Streams PerformancePART II – Real-life PC I/O hardware

PART I

Abstract approach

Based on Operating System Concepts by Avi Silberschatz, Peter Baer Galvin and Greg Gagne

http://codex.cs.yale.edu/avi/os-book/os7/

I/O Hardware

Incredible variety of I/O devices Common concepts

Port Bus (daisy chain or shared direct access)Controller (host adapter)

I/O instructions control devices Devices have addresses, used by

Direct I/O instructionsMemory-mapped I/O

A Typical PC Bus Structure

Device I/O Port Locations on PCs (partial)

Polling

Determines state of device command-readybusyError

Busy-wait cycle to wait for I/O from device

Interrupts CPU Interrupt-request line triggered by I/O device Interrupt handler receives interrupts Maskable to ignore or delay some interrupts

Interrupt vector to dispatch interrupt to correct handlerBased on prioritySome nonmaskable

Interrupt mechanism also used for exceptions

Interrupt-Driven I/O Cycle

Direct Memory Access

Used to avoid programmed I/O for large data movement

Requires DMA controller

Bypasses CPU to transfer data directly between I/O device and memory

Six Step Process to Perform DMA Transfer

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 dimensionsCharacter-stream or blockSequential or random-accessSharable or dedicatedSpeed of operationread-write, read only, or write only

A Kernel I/O Structure

Characteristics of I/O Devices

Block and Character Devices

Block devices include disk drivesCommands include read, write, seek Raw I/O or file-system accessMemory-mapped file access possible

Character devices include keyboards, mice, serial portsCommands include get, putLibraries layered on top allow line editing

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 availableUser interface, data copy (buffered I/O) Implemented via multi-threadingReturns quickly with count of bytes read or written

Blocking and Nonblocking I/O

Asynchronous - process runs while I/O executesDifficult to use I/O subsystem signals process when I/O

completed

Two I/O Methods

Synchronous Asynchronous

Kernel I/O Subsystem

SchedulingSome I/O request ordering via per-device queueSome OSs try fairness

Buffering - store data in memory while transferring between devicesTo cope with device speed mismatchTo cope with device transfer size mismatchTo maintain “copy semantics”

Kernel I/O Subsystem Caching - fast memory holding copy of data

Always just a copyKey to performance

Spooling - hold output for a device If device can serve only one request at a time i.e., Printing

Device reservation - provides exclusive access to a deviceSystem calls for allocation and deallocationWatch out for deadlock

Error Handling

OS can recover from disk read, device unavailable, transient write failures

Most return an error number or code when I/O request fails

System error logs hold problem reports

I/O Protection

User process may accidentally or purposefully attempt to disrupt normal operation via illegal I/O instructionsAll I/O instructions defined to be privileged I/O must be performed via system calls

Memory-mapped and I/O port memory locations must be protected too

I/O Requests to Hardware Operations Consider reading a file from disk for a

process: Determine device holding file Translate name to device representationPhysically read data from disk into bufferMake data available to requesting processReturn control to process

Life Cycle of An I/O Request

STREAMS

STREAM – a full-duplex communication channel between a user-level process and a device in Unix System V and beyond

A STREAM consists of:

- STREAM head interfaces with the user process

- driver end interfaces with the device- zero or more STREAM modules between them.

STREAMS

Each module contains a read queue and a write queue

Message passing is used to communicate between queues

The STREAMS Structure

Performance

I/O a major factor in system performance:

Demands CPU to execute device driver, kernel I/O code

Context switches due to interruptsData copyingNetwork traffic especially stressful

Intercomputer Communications

Improving Performance

Reduce number of context switches Reduce data copying Reduce interrupts by using large transfers,

smart controllers, polling Use DMA Balance CPU, memory, bus, and I/O

performance for highest throughput

PART II

Real-life PC I/O hardware

Based on various articles from Wikipediahttp://wikipedia.org

Drive interfaces

Allow to communicate with disk/optical drives

Short-ranged (with exceptions)

ATA (parallel)

Developed in late 80’s by Western Digital Integrated Drive Electronics (IDE) AT Attachment Packet Interface Bandwidth – up to 133 MB/s (UDMA-133)

SCSI

Developed in 1986 Special controller needed Bandwidth – from 1.5 MB/s (SCSI 1) to

640 MB/s (Ultra-640 SCSI)

SATA

Introduced in 2001 Hot-pluggable Natively supports command queueing Bandwidth – from 150 MB/s to 300 MB/s

Fibre Channel

Used for SANs Despite its name, runs on copper twisted-

pair cables too Topologies

Point-to-point (vide ordinary ATA)Arbitrated loop (vide token ring)Switched fabric (vide ethernet)

System buses

Used for communication with peripheral devices

Allow to extend functionality High bandwidth

ISA (Industry Standard Arch.)

Obsolete nowadays Bandwidth – from 4.77 MB/s (8-bit)

to 16.67 MB/s (16-bit)

PCI

Bandwidth – 133 MB/s (32-bit/33MHz) up to 533 MB/s (64-bit/66MHz)

Improvement – PCI-X

AGP

Designed to cope with big throughput required by video cards

Bandwidth – 266 MB/s (1x) up to2133 MB/s (8x)

PCI-X („Extended”)

Connector physically identical to PCI Do not confuse with PCI Express Bandwidth – up to 4.27 GB/s

PCI Express

Fast and robust Comes in many flavors – from 1x to 16x

(250 MB/s – 8GB/s)

External interfaces

Hot-pluggable (most of them) Allow further extension of functionality plus

additional convenience of use

Serial Port (RS-232)

Standard port used for serial communication, e.g.:Between a computer and a character deviceBetween two computers

Bandwidth – up to 115,200 bps

Parallel port

Transmits several bits in parallel (as the name suggests)

Omitted in modern computers Bandwidth – from 360 kBps to 2.5 MBps

PS/2

DIN/RS-232 replacement Not hot-pluggable Bandwidth – 40 kB/s Used for mice/keyboards

Universal Serial Bus

1.0 – about 0.2 MB/s 1.1 – 1.5 MB/s („Full-speed”) 2.0 – 60 MB/s („High-speed”) Power supply over bus

IEEE 1394 (FireWire , i.Link )

Similar to USB Bandwidth – up to 50 MB/s 1394b – 100 MB/s – 400 MB/s

The End

Any questions?