1. 1 Module 12: I/O Systems n I/O hardware n Application I/O Interface n Kernel I/O Subsystem n Transforming I/O Requests to Hardware Operations n Performance

2. 2 I/O Hardware n Incredible variety of I/O devices n Common concepts u Port u Bus (daisy chain or shared direct access) u Controller (host adapter) n I/O instructions control devices n Devices have addresses, used by u Direct I/O instructions u Memory-mapped I/O

3. 3 Polling n Determines state of device u command-ready u busy u error n Busy-wait cycle to wait for I/O from device

4. 4 Interrupts n CPU Interrupt request line triggered by I/O device n Interrupt handler receives interrupts n Maskable to ignore or delay some interrupts n Interrupt vector to dispatch interrupt to correct handler u Based on priority u Some unmaskable n Interrupt mechanism also used for exceptions

5. 5 Interrupt-drive I/O Cycle

6. 6 Direct Memory Access n Used to avoid programmed I/O for large data movement n Requires DMA controller n Bypasses CPU to transfer data directly between I/O device and memory

7. 7 Six step process to perform DMA transfer

8. 8 Application I/O Interface n I/O system calls encapsulate device behaviors in generic classes n Device-driver layer hides differences among I/O controllers from kernel n Devices vary in many dimensions u Character-stream or block u Sequential or random-access u Sharable or dedicated u Speed of operation u read-write, read only, or write only

9. 9 Block and Character Devices n Block devices include disk drives u Commands include read, write, seek u Raw I/O or file-system access u Memory-mapped file access possible n Character devices include keyboards, mice, serial ports  Commands include get, put u Libraries layered on top allow line editing

10. 10 Network Devices n Varying enough from block and character to have own interface n Unix and Windows/NT include socket interface u Separates network protocol from network operation  Includes select functionality n Approaches vary widely (pipes, FIFOs, streams, queues, mailboxes)

11. 11 Clocks and Timers n Provide current time, elapsed time, timer n if programmable interval time used for timings, periodic interrupts ioctl (on UNIX) covers odd aspects of I/O such as clocks and timers

12. 12 Blocking and Nonblocking I/O n Blocking - process suspended until I/O completed u Easy to use and understand u Insufficient for some needs n Nonblocking - I/O call returns as much as available u User interface, data copy (buffered I/O) u Implemented via multi-threading u Returns quickly with count of bytes read or written n Asynchronous - process runs while I/O executes u Difficult to use u I/O subsystem signals process when I/O completed

13. 13 Kernel I/O Subsystem n Scheduling u Some I/O request ordering via per-device queue u Some OSs try fairness n Buffering - store data in memory while transferring between devices u To cope with device speed mismatch u To cope with device transfer size mismatch u To maintain “copy semantics”

14. 14 Kernel I/O Subsystem n Caching - fast memory holding copy of data u Always just a copy u Key to performance n Spooling - hold output for a device u If device can serve only one request at a time u i.e., Printing n Device reservation - provides exclusive access to a device u System calls for allocation and deallocation u Watch out for deadlock

15. 15 Error Handling n OS can recover from disk read, device unavailable, transient write failures n Most return an error number or code when I/O request fails n System error logs hold problem reports

16. 16 Kernel Data Structures n Kernel keeps state info for I/O components, including open file tables, network connections, character device state n Many, many complex data structures to track buffers, memory allocation, “dirty” blocks n Some use object-oriented methods and message passing to implement I/O

17. 17 I/O Requests to Hardware Operations n Consider reading a file from disk for a process u Determine device holding file u Translate name to device representation u Physically read data from disk into buffer u Make data available to requesting process u Return control to process

18. 18 Life Cycle of an I/O Request

19. 19 Example: Disk Structure n Disk drives are addressed as large 1-dimensional arrays of logical blocks, where the logical block is the smallest unit of transfer. n The 1-dimensional array of logical blocks is mapped into the sectors of the disk sequentially. u Sector 0 is the first sector of the first track on the outermost cylinder. u Mapping proceeds in order through that track, then the rest of the tracks in that cylinder, and then through the rest of the cylinders from outermost to innermost.

20. 20 Disk Management n Low-level formatting, or physical formatting — Dividing a disk into sectors that the disk controller can read and write. n To use a disk to hold files, the operating system still needs to record its own data structures on the disk. u Partition the disk into one or more groups of cylinders. u Logical formatting or “making a file system”. n Boot block initializes system. u The bootstrap is stored in ROM. u Bootstrap loader program. n Methods such as sector sparing used to handle bad blocks or interleaving to handle different disk hardware designs

21. 21 Disk Performance n The operating system is responsible for using hardware efficiently — for the disk drives, this means having a fast access time and disk bandwidth. n Access time has three major components u Seek time is the time for the disk are to move the heads to the cylinder containing the desired sector. u Rotational latency is the additional time waiting for the disk to rotate the desired sector to the disk head. u Transfer delay is the time that takes to read the actual data from the disk and put it in memory

22. 22 Disk Performance n Because seek time is the largest of the delays, attempt to minimize it n Seek time  seek distance n Disk bandwidth is the total number of bytes transferred, divided by the total time between the first request for service and the completion of the last transfer.

23. 23 Speed n Bandwidth is measured in bytes per second. u Sustained bandwidth – average data rate during a large transfer; # of bytes/transfer time. Data rate when the data stream is actually flowing. u Effective bandwidth – average over the entire I/O time, including seek: it is the overall data rate of the device n Access latency – amount of time needed to locate data. u Access time for a disk – move the arm to the selected cylinder and wait for the rotational latency; < 35 milliseconds (about a thousand times faster than tape)

24. 24 Performance n I/O a major factor in system performance u Demands CPU to execute device driver, kernel I/O code u Context switches due to interrupts u Data copying u Network traffic especially stressful

25. 25 Intercomputer communications

26. 26 Improving Performance n Reduce number of context switches n Reduce data copying n Reduce interrupts by using large transfers, smart controllers, polling n Use DMA n Balance CPU, memory, bus, and I/O performance for highest throughput