1. I/O Systems

2. Outline I/O Hardware Application I/O Interface Kernel I/O Subsystem
Transforming I/O Requests to Hardware Operations
Streams
Performance

3. I/O Hardware There is incredible variety of I/O devices
There are shared common concepts
Device communicates with system using a connection point
Port
Bus (daisy chain or shared direct access)
Controller (host adapter)
Handshake between CPU and device controller done with polling/interrupts
Offloading to DMA for large I/O
Device-registers (control, state, data) accessible via the controller of device allow a computer to control a device
I/O instructions control devices
Devices have addresses, used by
Direct I/O instructions
Memory-mapped I/O

4. A Typical PC Bus Structure

5. Device I/O Port Locations on PCs (partial)

6. Polling Computer system determines state of device using
Interrupts
Polling determines state of device using certain commands
command-ready bit
Busy bit
Error bit
Command register
Data in/out registers
Busy-wait cycle to wait for I/O from device

7. Interrupts
CPU’s Interrupt Request Line is triggered by I/O device, and CPU checks after each instruction, and device sets interrupt address
Device raises interrupt
CPU catches interrupt and dispatches it
Handler clears the interrupt
Interrupt handler receives interrupts
Maskable to ignore or delay some interrupts
Interrupt vector to dispatch interrupt to correct handler
Based on priority
Some unmaskable
Chaining of handlers
Interrupt mechanism also used for exceptions

8. Interrupt-Driven I/O Cycle

9. Intel Pentium Processor Event-Vector Table

10. 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
CPU writes DMA control block
DMA controller interacts with device using
DMA-request raised by device to signal the DMA controller that device has data
DMA-acknowledge raised by DMA to tell device to put data on data bus

11. Six Step Process to Perform DMA Transfer

12. Application I/O Interface
I/O system calls encapsulate device behaviors in generic classes
Device-driver layer hides differences among I/O device 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

13. A Kernel I/O Structure

14. Characteristics of I/O Devices

15. Block and Character Devices
Block devices include disk drives
Commands include read, write, seek
Raw I/O (device is viewed as a linear array of logical blocks) or file-system access
Memory-mapped file access possible
Executing a program can be done with memory-mapping of the executable to VM
Character devices include keyboards, mice, serial ports
Commands include get, put
Libraries layered on top allow line editing

16. Network Devices
Varying enough from block and character to have own interface
Unix and Windows NT/9x/2000 include socket interface
Similar to electric sockets
Local and remote sockets and connecting them with addresses
Separates network protocol from network operation
Includes select functionality that manage a set of sockets and inform the CPU which socket have data or are ready to accept more data
Approaches vary widely
pipes, FIFOs, streams, queues, mailboxes

17. Clocks and Timers Provide current time, elapsed time, timer
Programmable Interval Timer can be used for implementing timers (via periodic interrupts)
OS can implement many timers using a single Programmable timer
ioctl (on UNIX) covers odd aspects of I/O such as clocks and timers

18. 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 data 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 use
I/O subsystem signals process when I/O completed

19. Kernel I/O Subsystem Scheduling to improve performance
Some I/O request ordering via per-device queue
Some OSs try fairness
Buffering - store data in memory while transferring between devices
To cope with device speed mismatch
To cope with device transfer size mismatch
To maintain “copy semantics”
To allow programmer to overwrite user I/O buffer as soon as I/O command returns, without having to worry about when data have been output to the device
Can also be implemented with VM and copy-on-write

20. Sun Enterprise 6000 Device-Transfer Rates

21. Kernel I/O Subsystem Caching - fast memory holding copy of data
Always just a copy
Key to performance
Sometimes can be used together with buffering
Spooling - hold output for a device
If device can serve only one request at a time
i.e., printers, tape drives, etc
Device reservation - provides exclusive access to a device
System calls for allocation and deallocation of devices
Watch out for deadlocks

22. 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

23. 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, etc
Some use object-oriented methods and message passing to implement I/O

24. UNIX I/O Kernel Structure

25. From 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

26. Life Cycle of An I/O Request

27. STREAMS
STREAM – a full-duplex communication channel between a user-level process and a device
A STREAM consists of:
STREAM head that interfaces with the user process using read/write or get/put message in a synchronous manner
driver end that interfaces with the device
and zero or more STREAM modules between them, that communicate via message passing in asynchronous manner
Each module contains a read queue and a write queue
Message passing is used to communicate between queues, and may implement flow control
STREAMs provide a modular and incremental approach to developing device drivers and protocols

28. The STREAMS Structure

29. Performance I/O is a major factor in system performance:
Demands CPU to execute device driver, kernel I/O code
Context switches due to interrupts
Data copying
Network traffic especially stressful

30. Intercomputer Communications

31. 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
But where should I/O functionality be implemented?
User process?
Kernel?
Device driver?
Device controller?

32. Device-Functionality Progression