Real-Time Operating Systems (Introduction)

1
Real-Time Operating Systems(Introduction)

• M. KargahiSchool of ECE University of Tehran

2
Real-Time Systems

• A Real-Time System is a system whose
  specification include
• Logically correctness of computations
• Temporally correctness (timeliness) of
  computations
• i.e., High predictability
• Sometimes called Reactive Systems
• Deadline
• Relative deadline ( hour / min / sec / msec /
  ?sec )
• Some applications
• Air Traffic Control (ATC)
• Interacts with radar, M.M. sub-system, operator,
  
• Real-time databases
• Multimedia and streaming applications
• Electronic games
• Mobile applications

3
Real-Time Embedded Systems

• Requirements
• Environmental size, power (heat), weight, and
  radiation-hardened
• Performance responsiveness, predictability
• Economic cost, time-to-market
• Consequence safety, fault-tolerance, security
• Solutions
• Design process (specification, development,
  testing, etc.)
• Hardware (processor, memory, I/O, bus, etc.)
• Software (OS, libraries, applications, GUI, etc.)
• Tool chain (analysis, compiling, debugging,
  integration, etc.)

4
SW Development for RT EM Systems

• Requirements
• Multitasking for concurrent events (real-world
  events occur in parallel)
• Portability
• Software abstraction and modular design
• Verifiability, reusability, maintainability
• Controlling timing with a good granularity
• Sharing resources and dominating resource
  constraints
• Scheduling tasks, messages, and I/O
• Therefore, we need real-time (embedded) operating
  systems
• Moreover, we need development, monitoring, and
  test environments

5
Disciplines that Impact on Real-Time Systems
Engineering
6
Basic concepts and terminologies

• Job/Service/Message/Packet Each unit of work
  that is scheduled and executed by the system. (to
  be allocated processor time and other resources)
• Task A set of related jobs that jointly provide
  some system function.
• Release time The instant of time at which the
  job becomes available for execution.
• Deadline The instant of time by which the job
  execution is required to be completed.
• Response time The length of time from the event
  sense time of the job to the instant when
  actuation is done (output completes).

7
Real-time systems classification

• Conventional classification
• Hard Real-Time (HRT)
• Soft Real-Time (SRT)
• Firm Real-Time (FRT)
• Criteria for the classification
• Functional criticality of jobs
• Usefulness of late results (Usefulness/Utility
  function)
• Tardiness Max (0 ,Completion-time Deadline)
• Deterministic or probabilistic nature of
  constraints

8
Real-time systems classification

• HRT systems
• Critical systems (ATC, Anti-Lock Braking)
• Late results have negative values and usually
  result in catastrophic consequences ( if
  tardinessgt0 ?F(tardiness)lt0 )
• Validation for meeting the timing constraints is
  required
• Almost always only deterministic nature is valid
  for HRT systems (Weakly HRT Systems)
• Systems where m out of k deadlines have to be met
• Feedback control systems, in which the control
  becomes unstable with too many missed control
  cycles
• Suitable for dealing with other failures, e.g.,
  Electro Magnetic Interference EMI
• There may be no advantage in early completing a
  job with a hard deadline
• It is often advantageous, sometimes even
  essential, to keep jitters in the response time
  of a stream of jobs small

9
Real-time systems classification

• SRT systems
• Non-critical systems (Multimedia, Virtual
  Reality, Electronic Games, Online Transaction
  Systems, Telephone Switches)
• Usefulness functions specify the value of the
  late results (F(x)?1)
• F(tardiness)?0 for SRT systems is a decreasing
  function
• A system is SOFTER than the other if F(tardiness)
  decreases at a slower rate for which
• The timing requirements are often specified in
  probabilistic terms
• Usually, having a small average response time and
  high throughput is more important than meeting
  all deadlines

10
Real-time systems classification

• FRT systems
• Non-critical systems (Forecasting systems,
  Teleconferencing, Reporting a News, Doing a
  Homework)
• Late results have zero value (i.e., if
  tardinessgt0 ? F(tardiness)0 )
• The timing requirements are often specified in
  probabilistic terms

11
Hard Real-Time Service Utility
12
Soft Real-Time Service Utility
13
Best Effort Service Utility
14
Isochronal Hard Real-Time Utility
15
Isochronal Soft Real-Time Utility
16
Anytime Utility Curve
17
A Reference Model

• Workload model Describes the application
• Resource model Describes the system resources
  available to the applications
• Resource-to-workload allocation algorithms
  Mainly is applied using the OS
• Processors Resources
• Processors Servers Active resources,
  Transmission links.
• Resources (Passive) Memory, Sequence numbers,
  Mutexes, Database locks
• Example Sliding window
• Job Message transfer
• Processor Data link
• Resource Sequence numbers (multiple instance
  reusable resources)

18
Workload Model

• Ji job number i
• ri release time
• ei execution time
• di absolute deadline
• Di relative deadline
• ri-, ri jittered release time
• ei-, ei jittered execution time

19
Task Execution Time

• Estimation of the execution time depends on
• Source code and the complexity of the job
• Compiler non-unique mapping of source to object
  code
• Machine architecture
• No. of registers
• The size and organization of the cache
• Clock rate
• Memory refresh time affects on the effective
  access time
• pipeline,
• OS task scheduling but not on how the job is
  scheduled, memory management, interrupt handling
  overhead,

20
Workload Model

• Analysis of straight line source code
• L1 a bc
• L2 b de
• L3 d a-f
• L1 can be translated to
• L1.1 Get the address of c
• L1.2 Load c
• L1.3 Get the address of b
• L1.4 Load b
• L1.5 Multiply
• L1.6 Store into a
• When does the execution time of L1 become
  ?i1..6Texec(L1.i)?
• Even with no pipeline no interrupts, multiply
  time depends on data

21
Workload Model

• L4 while (p) do
• L5 Q1
• L6 Q2
• L7 Q3
• L8 end_while
• L9 if B1 then S1
• else if B2 then S2
• else if B3 then S3
• else S4
• end_if
• If B1 is true then T(B1)T(S1)T(JMP).
• If (not B1).B2 then T(B1)T(B2)T(S2)T(JMP).
• for may be better than while in writing
  programs!
• Using OO languages may not be a good idea!
• What if interrupts?

22
Scheduling (Resource Allocation Model)

• A feasible schedule
• A valid schedule by which every job completes by
  its deadline
• A HRT scheduling algorithm is optimal
• If using the algorithm, the scheduler always
  produces a feasible schedule if the given set of
  jobs has any feasible schedules
• Interaction among schedulers
• A system typically has a hierarchy of schedulers
• Logical resources should be scheduled on physical
  resources, e.g., scheduling a DB and its locks
• Some servers may be scheduled using the OS, while
  the server schedules its client using a different
  scheduler
• Periodic tasks and the poller may be scheduled by
  the OS, while the poller itself schedules the
  aperiodic jobs

23
More considerations

• Fault
• Transient
• Intermittent
• Permanent 1/30th of the transient faults
• Redundancy
• Spatial
• Time
• Power
• Transistor-level
• Gate-level
• Architecture-level
• OS-level
• Application-level

24
Performance Measures

• Some traditional performance measures
• Maximum and average tardiness
• Maximum and average lateness
• Latenesscompletion-time deadline
• Unlike tardiness, lateness can be negative
• Maximum and average response time
• Miss rate
• Loss rate
• Invalid ratemiss rate loss rate