here - SafeTTy Systems Ltd
The Engineering of
Reliable Embedded Systems
LPC1769 edition
Michael J. Pont
This document includes extracts from the book:
Pont, M.J. (2014) “The Engineering of Reliable Embedded
Systems: LPC1769 edition”, Published by SafeTTy Systems Ltd.
ISBN: 978-0-9930355-0-0.
Last updated: 21 November 2014
This document may be freely distributed
(provided that the file is not altered in any way)
2014
Published by SafeTTy Systems Ltd
www.SafeTTy.net
First published 2014
First printing November 2014
Copyright В© 2014 by SafeTTy Systems Ltd
The right of Michael J. Pont to be identified as Author of this work has been
asserted by him in accordance with the Copyright, Designs and Patents Act 1988.
ISBN 978-0-9930355-0-0
All rights reserved; no part of this publication may be reproduced, stored in a
retrieval system, or transmitted in any form or by any means, electronic,
mechanical, photocopying, recording, or otherwise without the prior written
permission of the publishers. This book may not be lent, resold, hired out or
otherwise disposed of in any form of binding or cover other than that in which it is
published, without the prior consent of the publishers.
Trademarks
MoniTTorВ® is a registered trademark of SafeTTy Systems Ltd.
PredicTTorВ® is a registered trademark of SafeTTy Systems Ltd.
ReliabiliTTyВ® is a registered trademark of SafeTTy Systems Ltd.
SafeTTy SystemsВ® is a registered trademark of SafeTTy Systems Ltd.
ARMВ® is a registered trademark of ARM Limited
NXPВ® is a registered trademark of NXP Semiconductors
All other trademarks acknowledged.
British Library Cataloguing in Publication Data
A catalogue record for this book is available from the British Library.
This book is dedicated to Benjamin, Timothy, Rowena, Jonathan and Eliza.
Contents
Definitions ................................................................................................ xv
Acronyms and abbreviations .................................................................... xvii
Reference designs ..................................................................................... xix
International standards and guidelines ...................................................... xxi
Preface ................................................................................................... xxiii
a. What is a “reliable embedded system”? .............................................................. xxiii
b. Who needs reliable embedded systems? ............................................................ xxiii
c. Why work with “time-triggered” systems? .......................................................... xxiv
d. How does this book relate to international safety standards? ............................. xxv
e. What programming language is used? ................................................................. xxv
f. Is the source code “freeware”? ............................................................................. xxv
g. How does this book relate to other books in the “ERES” series? ........................ xxvi
h. What processor hardware is used in this book? .................................................. xxvi
i. How does this book relate to “PTTES”? ............................................................... xxvii
j. Is there anyone that you’d like to thank? ............................................................ xxvii
PART ONE: INTRODUCTION ......................................................................... 1
CHAPTER 1: Introduction ............................................................................. 3
1.1. Introduction .......................................................................................................... 3
1.2. Single-program, real-time embedded systems ..................................................... 4
1.3. TT vs. ET architectures .......................................................................................... 6
1.4. Modelling system timing characteristics .............................................................. 7
1.5. Working with “TTC” schedulers ............................................................................ 9
1.6. Supporting task pre-emption .............................................................................. 11
1.7. Different system modes...................................................................................... 11
1.8. A “Model-Build-Monitor” methodology ............................................................. 12
1.9. How can we avoid Uncontrolled System Failures? ............................................. 14
1.10. Conclusions ....................................................................................................... 16
CHAPTER 2: Creating a simple TTC scheduler .............................................. 17
2.1. Introduction ........................................................................................................ 17
2.2. A first TTC scheduler (TTRD02a) ......................................................................... 21
2.3. The scheduler data structure and task array ...................................................... 21
2.4. The �Init’ function................................................................................................ 21
2.5. The �Update’ function ......................................................................................... 23
2.6. The �Add Task’ function ...................................................................................... 24
2.7. The �Dispatcher’ .................................................................................................. 24
2.8. The �Start’ function ............................................................................................. 26
2.9. The �sleep’ function ............................................................................................ 26
2.10. Where is the “Delete Task” function? .............................................................. 27
2.11. Watchdog timer support .................................................................................. 28
2.12. Choice of watchdog timer settings ................................................................... 29
vii
2.13. The �Heartbeat’ task (with error reporting) ...................................................... 30
2.14. Detecting system overloads (TTRD02b) ............................................................ 31
2.15. Example: Injected (transitory) task overrun (TTRD02b) ................................... 32
2.16. Task overruns may not always be “A Bad Thing” ............................................. 33
2.17. Porting the scheduler (TTRD02c) ...................................................................... 33
2.18. Conclusions ....................................................................................................... 34
2.19. Code listings (TTRD02a) .................................................................................... 35
2.20. Code listings (TTRD02b) .................................................................................... 60
2.21. Code listings (TTRD02c) .................................................................................... 61
CHAPTER 3: Initial case study ..................................................................... 63
3.1. Introduction ........................................................................................................ 63
3.2. The focus of this case study ................................................................................ 63
3.3. The purpose of this case study ........................................................................... 63
3.4. A summary of the required system operation .................................................... 65
3.5. The system architecture ..................................................................................... 65
3.6. The system states ............................................................................................... 66
3.7. Implementation platform for the prototype ...................................................... 66
3.8. The “system” task ............................................................................................... 68
3.9. The “selector dial” task ....................................................................................... 68
3.10. The “start switch” task ...................................................................................... 69
3.11. The “door lock” task ......................................................................................... 69
3.12. The “water valve” task ...................................................................................... 69
3.13. The “detergent hatch” task .............................................................................. 69
3.14. The “water level” task....................................................................................... 69
3.15. The “water heater” task ................................................................................... 69
3.16. The “water temperature” task.......................................................................... 69
3.17. The “drum motor” task ..................................................................................... 69
3.18. The “water pump” task ..................................................................................... 69
3.19. The “heartbeat” task ........................................................................................ 69
3.20. Communication between tasks ........................................................................ 70
3.21. Where do we go from here? ............................................................................. 71
3.22. Conclusions ....................................................................................................... 72
3.23. Code listings (TTRD03a) .................................................................................... 73
PART TWO: CREATING RELIABLE TTC DESIGNS ............................................ 91
CHAPTER 4: Modelling system timing characteristics................................... 93
4.1. Introduction ........................................................................................................ 93
4.2. Basic Tick Lists ..................................................................................................... 93
4.3. Determining the required tick interval ............................................................... 94
4.4. Working with “Short Tasks” ................................................................................ 95
4.5. The hyperperiod ................................................................................................. 95
4.6. Performing GCD and LCM calculations ............................................................... 96
4.7. Synchronous and asynchronous task sets .......................................................... 96
4.8. Modelling CPU loading........................................................................................ 96
4.9. Worked Example A: Determining the maximum CPU load ................................. 98
4.10. Worked Example A: Solution ............................................................................ 99
viii
4.11. Modelling task jitter ........................................................................................ 100
4.12. Worked Example B: Modelling task release jitter ........................................... 104
4.13. Worked Example B: Solution .......................................................................... 104
4.14. Modelling response times .............................................................................. 105
4.15. Worked Example C: An “emergency stop” interface ...................................... 107
4.16. Worked Example C: Solution .......................................................................... 110
4.17. Creating Tick Lists ........................................................................................... 111
4.18. Conclusions ..................................................................................................... 112
CHAPTER 5: Obtaining data for system models ......................................... 113
5.1. Introduction ...................................................................................................... 113
5.2. The importance of WCET / BCET information................................................... 113
5.3. Challenges with WCET / BCET measurements .................................................. 114
5.4. Instrumenting a TTC scheduler: WCET-BCET (TTRD05a)................................... 116
5.5. Example: An injected task overrun (TTRD05b) ................................................. 117
5.6. Obtaining jitter measurements: Tick jitter (TTRD05c) ...................................... 117
5.7. Example: The impact of idle mode on a TTC scheduler .................................... 117
5.8. Obtaining jitter measurements: Task jitter (TTRD05d) ..................................... 118
5.9. Example: The impact of task order on a TTC scheduler .................................... 119
5.10. Traditional ways of obtaining task timing information ................................... 121
5.11. Other ways of obtaining task timing information ........................................... 121
5.12. Generating a Tick List on an embedded platform (TTRD05e) ......................... 121
5.13. Generating a Tick List on a PC or similar platform .......................................... 123
5.14. Adjusting task offsets ...................................................................................... 123
5.15. Conclusions ..................................................................................................... 124
5.16. Code listing (TTRD05a) .................................................................................... 125
5.17. Code listing (TTRD05b) ................................................................................... 127
5.18. Code listing (TTRD05c) .................................................................................... 131
5.19. Code listing (TTRD05d) ................................................................................... 136
5.20. Code listing (TTRD05e).................................................................................... 140
CHAPTER 6: Timing considerations when designing tasks .......................... 145
6.1. Introduction ...................................................................................................... 145
6.2. Design goal: “Short Tasks” ................................................................................ 146
6.3. The need for multi-stage tasks ......................................................................... 146
6.4. Example: Measuring liquid flow rates .............................................................. 148
6.5. Example: Buffering output data........................................................................ 150
6.6. Example: DMA-supported outputs ................................................................... 152
6.7. The need for timeout mechanisms ................................................................... 152
6.8. Example: Loop timeouts ................................................................................... 156
6.9. Example: Hardware timeout ............................................................................. 156
6.10. Handling large / frequent data inputs ............................................................ 157
6.11. Example: Buffered input ................................................................................. 157
6.12. Example: DMA input ....................................................................................... 157
6.13. Example: Multi-core input (“Smart” buffering) .............................................. 158
6.14. Execution-time balancing in TTC designs (task level) ..................................... 158
6.15. Execution-time balancing in TTC designs (within tasks) ................................. 159
ix
6.16. Execution-time balancing in TTH / TTP designs .............................................. 161
6.17. Example: Execution-time balancing at an architecture level .......................... 161
6.18. Example: Manual execution-time balancing................................................... 162
6.19. Example: Sandwich delays for execution-time balancing ............................... 163
6.20. Appropriate use of Sandwich Delays .............................................................. 165
6.21. Conclusions ..................................................................................................... 166
CHAPTER 7: Multi-mode systems.............................................................. 167
7.1. Introduction ...................................................................................................... 167
7.2. What does it mean to change the system mode? ............................................ 167
7.3. The timing of mode changes............................................................................. 168
7.4. Implementing effective multi-mode designs .................................................... 168
7.5. Mode changes vs. state changes ...................................................................... 169
7.6. Key components ............................................................................................... 169
7.7. System settings ................................................................................................. 170
7.8. Design example (TTRD07a) ............................................................................... 170
7.9. Supporting “graceful degradation” ................................................................... 172
7.10. Design example (TTRD07b) ............................................................................. 172
7.11. Conclusions ..................................................................................................... 173
7.12. Code listings (TTRD07a) .................................................................................. 174
7.13. Code listings (TTRD07b) .................................................................................. 178
CHAPTER 8: Task Contracts (Resource Barriers) ......................................... 185
8.1. Introduction ...................................................................................................... 185
8.2. Origins of “Contracts” in software development.............................................. 186
8.3. What do we mean by a “Task Contract”? ......................................................... 186
8.4. Numerical example ........................................................................................... 187
8.5. Control example ............................................................................................... 187
8.6. Timing is part of the Task Contract ................................................................... 188
8.7. Implementing Task Contracts (overview) ......................................................... 189
8.8. Implementing Task Contracts (timing checks) .................................................. 190
8.9. Implementing Task Contracts (checking peripherals) ....................................... 190
8.10. Example: Feeding the WDT ............................................................................. 193
8.11. One task per peripheral .................................................................................. 193
8.12. What about shared data? ............................................................................... 195
8.13. Implementing Task Contracts (protecting data transfers) .............................. 195
8.14. Example: Switch and LED task pair ................................................................. 196
8.15. How can we detect corruption of the scheduler data? .................................. 196
8.16. Making use of the MPU .................................................................................. 196
8.17. What do we do if our resource barrier detects a fault? ................................. 197
8.18. Injecting faults ................................................................................................ 197
8.19. Designing and implementing Backup Tasks .................................................... 197
8.20. The complete scheduler ................................................................................. 198
8.21. Task Contracts and international standards ................................................... 198
8.22. Conclusions ..................................................................................................... 199
8.23. Code listings (TTRD08a) .................................................................................. 200
x
CHAPTER 9: Task Contracts (Time Barriers)............................................... 215
9.1. Introduction ...................................................................................................... 215
9.2. An overview of the system architecture ........................................................... 216
9.3. System operation .............................................................................................. 216
9.4. Implementation example: TTC scheduler with MoniTTor (TTRD09a)............... 218
9.5. External MoniTTor solutions ............................................................................. 218
9.6. Alternatives to MoniTTor.................................................................................. 218
9.7. International standards .................................................................................... 219
9.8. Conclusions ....................................................................................................... 219
9.9. Code listings (TTRD09a) .................................................................................... 219
CHAPTER 10: Monitoring task execution sequences.................................. 235
10.1. Introduction .................................................................................................... 235
10.2. The goal of task-sequence monitoring ........................................................... 235
10.3. A dynamic scheduler and a static task-sequence list ...................................... 236
10.4. Synchronous vs. asynchronous task sets revisited ......................................... 236
10.5. The Task Sequence Initialisation Period (TSIP) ............................................... 237
10.6. Worked example............................................................................................. 239
10.7. Solution ........................................................................................................... 239
10.8. Creating the Task-Sequence Representation (TSR) ........................................ 240
10.9. Where should we store the TSR? .................................................................... 240
10.10. Storing the TSR in internal memory (TTRD10a) ............................................ 240
10.11. Protecting the TSR using the MPU (TTRD10b) .............................................. 240
10.12. Storing the TSR in an external memory device (TTRD10c) ........................... 241
10.13. Monitoring task sequences and execution time (TTRD10d) ......................... 241
10.14. Conclusions ................................................................................................... 241
10.15. Code listings (TTRD10a) ................................................................................ 241
10.16. Code listings (TTRD10b) ................................................................................ 258
10.17. Code listings (TTRD10c) ................................................................................ 272
PART THREE: CREATING RELIABLE TTH AND TTP DESIGNS ......................... 287
CHAPTER 11: Supporting task pre-emption .............................................. 289
11.1. Introduction .................................................................................................... 289
11.2. Implementing a TTH scheduler ....................................................................... 291
11.3. Key features of a TTH scheduler ..................................................................... 292
11.4. TTH example: Emergency stop (TTRD11a) ...................................................... 293
11.5. TTH example: Medical alarm in compliance with IEC60601-1-8..................... 294
11.6. TTH example: Long pre-empting section (TTRD11b) ...................................... 295
11.7. From TTH to TTP (TTRD11c) ............................................................................ 296
11.8. Monitoring task execution times (TTRD11d) .................................................. 296
11.9. Use of watchdog timers in TTH and TTP designs ............................................ 299
11.10. Conclusions ................................................................................................... 300
11.11. Code listings .................................................................................................. 301
CHAPTER 12: Maximising temporal determinism ...................................... 303
12.1. Introduction .................................................................................................... 303
12.2. Jitter levels in TTH designs (TTRD12a) ............................................................ 303
xi
12.3. Reducing jitter in TTH designs (TTRD12b) ....................................................... 304
12.4. Shared resources and priority inversion in ET systems .................................. 305
12.5. The impact of PI on ET and TT designs............................................................ 307
12.6. Avoiding priority inversion in TTH / TTP systems ........................................... 307
12.7. A general need for code balancing in TTH / TTP designs ................................ 308
12.8. Do you need to balance the code in your system? ......................................... 309
12.9. Using code balancing to prevent priority inversion ........................................ 310
12.10. Monitoring task execution sequences (TTRD12c)......................................... 311
12.11. Conclusions ................................................................................................... 311
8.12. Code listings (TTRD12a) .................................................................................. 311
PART FOUR: COMPLETING THE SYSTEM .................................................... 327
CHAPTER 13: Working with System Contracts ........................................... 329
13.1. Introduction .................................................................................................... 329
13.2. What is a “System Contract”? ......................................................................... 330
13.3. Three computational blocks ........................................................................... 330
13.4. Generic POST operations ................................................................................ 331
13.5. Example: POST operations that meet IEC 60335 requirements ..................... 333
13.6. Checking the system configuration................................................................. 334
13.7. Example: Check the system configuration ...................................................... 335
13.8. Generic periodic checks (BISTs) ...................................................................... 336
13.9. Example: BISTs in compliance with IEC 60335 ................................................ 336
13.10. Additional periodic tests ............................................................................... 336
13.11. Example: Monitoring CPU temperature (TTRD13a) ...................................... 337
13.12. Shutting the system down in the event of faults .......................................... 337
13.13. Shutting the system down normally ............................................................. 338
13.14. Tasks and backup tasks ................................................................................. 338
13.15. Example: Design of a backup task ................................................................. 341
13.16. International standards ................................................................................ 341
13.17. Conclusions ................................................................................................... 342
8.18. Code listings (TTRD13a) .................................................................................. 342
CHAPTER 14: Recommended system architectures.................................... 359
14.1. Introduction .................................................................................................... 359
14.2. Can you use a TTC architecture?..................................................................... 359
14.3. You need to use a TTC architecture? .............................................................. 359
14.4. Modelling the system ..................................................................................... 360
14.5. Where should we store the task execution sequence data? .......................... 360
14.6. Towards a TT processor .................................................................................. 361
14.7. Recommended system architecture: TT1 ....................................................... 362
14.8. Recommended System architecture: TT2 ....................................................... 362
14.9. Internal monitoring vs independent monitoring ............................................ 363
14.10. The challenge of common-cause failures ..................................................... 364
14.11. Recommended system architecture: TT3 ..................................................... 364
14.12. Alternative implementations of TT3 ............................................................. 365
13.13. Fault-injection tests ...................................................................................... 365
14.14. Diagnostic coverage of the different architectures ...................................... 366
xii
14.15. Conclusions ................................................................................................... 370
CHAPTER 15: Revisiting the case study ..................................................... 371
15.1. Introduction .................................................................................................... 371
15.2. Reviewing the initial design ............................................................................ 372
15.3. Three implementations .................................................................................. 372
15.4. Changes cf. initial version ............................................................................... 373
15.5. Results of fault injection on both designs ....................................................... 373
15.6. International safety standards ........................................................................ 374
15.7. Conclusions ..................................................................................................... 376
15.8. Code listings (TTRD15a) .................................................................................. 377
15.9. Code listings (TTRD15b) .................................................................................. 391
15.10. Code listings (TTRD15c) ................................................................................ 405
PART FIVE: CONCLUSIONS ....................................................................... 421
CHAPTER 16: Conclusions ........................................................................ 423
16.1. Introduction .................................................................................................... 423
16.2. Basic TT designs .............................................................................................. 424
16.3. Real-time TT designs using Contracts ............................................................. 425
16.4. From processor to distributed system ............................................................ 427
16.5. Conclusions ..................................................................................................... 428
APPENDICES............................................................................................ 429
APPENDIX 1: Processor selection ............................................................. 431
A1.1. Introduction .................................................................................................... 431
A1.2. General considerations................................................................................... 431
A1.3. Supporting a TT scheduler .............................................................................. 432
A1.4. Supporting an external TT monitoring system ............................................... 433
A1.5. Conclusions..................................................................................................... 433
APPENDIX 2: LPC1769 test platform ......................................................... 435
A2.1. Introduction .................................................................................................... 435
A2.2. Heartbeat example (TTRD02a) ....................................................................... 435
A2.3. Debugging the system .................................................................................... 435
A2.4. Using the USB interface .................................................................................. 436
A1.5. Conclusions..................................................................................................... 437
Full list of references and related publications ......................................... 437
Index ...................................................................................................... 451
xiii
Definitions
An Uncontrolled System Failure means that the system has not detected a
System Fault correctly or – having detected such a fault – has not
executed a Controlled System Failure correctly, with the consequence
that significant System Damage may be caused.
A Controlled System Failure means that – having correctly detected a
System Fault – a reset is performed, after which the system enters a
normal operating mode, or a Limp-Home Mode, or a Fail-Silent Mode.
A Controlled System Failure may proceed in stages. For example, after
a System Fault is detected, a normal reset may be performed; if another
System Fault is detected within a pre-determined interval (e.g. 1 hour),
the system may then enter a Limp-Home Mode. Depending on the
nature of the fault, the sequence may vary: for example, the system may
move immediately to a Fail-Silent Mode in some circumstances.
A Limp-Home Mode means a pre-determined dynamic mode in which –
while the system is not fully operational – a core subset of the system
requirements are being met, and little or no System Damage is being
caused. In many cases, the system will enter a Limp-Home Mode on a
temporary basis (for example, while attempts are made to bring a
damaged road vehicle to rest in a location at the side of a motorway),
before it enters a Fail-Silent Mode.
A Fail-Silent Mode means a pre-determined static mode in which the
system has been shut down in such a way that it will cause little or no
System Damage. In most cases, it is expected that intervention by a
qualified individual (e.g. a Service Engineer) may be required to re-start
the system.
System Damage results from action by the system that is not in accordance
with the system requirements. System Damage may involve loss of life
or injury to users of the system, or to people in the vicinity of the
system, or loss of life or injury to other animals. System Damage may
involve direct or indirect financial losses. System Damage may involve
a wider environmental impact (such as an oil spill). System Damage
may involve more general damage (for example, through incorrect
activation of a building sprinkler system).
A System Fault means a Hardware Fault and / or a Software Fault.
xv
A Software Fault means a manifestation of a Software Error or a Deliberate
Software Change.
A Hardware Fault means a manifestation of a Hardware Error, or a
Deliberate Hardware Change, or the result of physical damage. Physical
damage may arise – for example – from a broken connection, or from
the impact of electromagnetic interference (EMI), radiation, vibration or
humidity.
A Deliberate Software Change means an intentional change to the
implementation of any part of the System Software that occurs as a
result of a “computer virus” or any other form of malicious interference.
A Software Error means a mistake in the requirements, design, or
implementation (that is, programming) of any part of the System
Software.
A Deliberate Hardware Change means an intentional change to the
implementation of any part of the System Hardware that occurs as a
result of any form of malicious interference.
A Hardware Error means a mistake in the requirements, design, or
implementation of any part of the System Hardware.
System Software means all of the software in the system, including tasks,
scheduler, any support libraries and “startup” code.
System Hardware means all of the computing and related hardware in the
system, including any processing devices (such as microcontrollers,
microprocessors, FPGAs, DSPs and similar items), plus associated
peripherals (e.g. memory components) and any devices under control of
the computing devices (e.g. actuators), or providing information used by
these devices (e.g. sensors, communication links).
xvi
Acronyms and abbreviations
ASIL
BCET
CAN
CBD
CLPD
CMSIS
COTS
CPU
DMA
DVS
ECU
EMI
ET
FAP
FFI
FPGA
FS
FSR
MC
MCU
MMU
MPU
PMU
PSU
PTTES
RMA
SIL
SoC
STA
TG
TSIP
TT
TTC
TTH
TTP
TTRD
WCET
WDT
Automotive Safety Integrity Level
Best-Case Execution Time
Controller Area Network
Contract-Based Design
Complex Programmable Logic Device
Cortex Microcontroller Software Interface Standard
Commercial �Off The Shelf’
Central Processor Unit
Deadline Monotonic Analysis
Dynamic Voltage Scaling
Electronic Control Unit
Electromagnetic Interference
Event Triggered
Failure Assertion Programming
Freedom From Interference
Field Programmable Gate Array
Functional Safety
Functional Safety Requirement
Mixed Criticality
Microcontroller (Unit)
Memory Management Unit
Memory Protection Unit
Predictive Monitoring Unit
Power Supply Unit
Patterns for Time-Triggered Embedded Systems
Rate Monotonic Analysis
Safety Integrity Level
System on Chip
Static Timing Analysis
Task Guardian
Task Sequence Initialisation Period
Time Triggered
Time-Triggered Co-operative
Time-Triggered Hybrid
Time-Triggered Pre-emptive
Time-Triggered Reference Design
Worst-Case Execution Time
Watchdog Timer
xvii
1
Reference designs
TTRD02a
TTRD02b
TTRD02c
TTRD03a
TTRD05a
TTRD05b
TTRD05c
TTRD05d
TTRD05e
TTRD07a
TTRD07b
TTRD08a
TTRD09a
TTRD10a
TTRD10b
TTRD10c
TTRD10d
TTRD11a
TTRD11b
TTRD11c
TTRD12a
TTRD12b
TTRD12c
TTRD12d
TTRD12e
TTRD13a
TTRD15a
TTRD15a
TTRD15a
1
TTC scheduler with �Heartbeat’ error reporting
TTC scheduler with injected task overrun
TTC scheduler (porting example)
Simple TTC control unit for washing machine
Instrumented TTC scheduler (BCET and WCET)
Instrumented TTC scheduler with task overrun
Instrumented TTC scheduler (tick jitter)
Instrumented TTC scheduler (task jitter)
TTC Dry scheduler
TTC architecture with multiple (normal) operating modes
TTC architecture with “Normal”, “Limp Home”
and “Fail Silent” modes.
TTC-TC MPU scheduler
TTC MoniTTor architecture (internal)
TTC PredicTTor architecture (generic)
TTC PredicTTor architecture (MPU support)
TTC PredicTTor architecture (external EEPROM)
TTC MoniTTor / PredicTTor architecture
TTH scheduler with “emergency stop”
TTH scheduler with long pre-empting tasks
TTP scheduler (BCET / WCET monitoring)
Instrumented TTH scheduler (tick jitter)
TTH scheduler with reduced release jitter (idle task)
TTP scheduler with reduced release jitter (idle task)
TTP scheduler with PredicTTor
TTP scheduler with MoniTTor and PredicTTor
TTC scheduler with temperature monitoring
Washing machine controller (TT1)
Washing machine controller (TT2)
Washing machine controller (TT3)
See: http://www.safetty.net/downloads/reference-designs
xix
International standards and guidelines
Reference in text
DO-178C
Full reference
DO-178C: 2012
IEC 60335-1
IEC 60601-1-8
IEC 60730-1
IEC 61508
IEC 62304
IEC 60335-1:2010 + A1: 2013
IEC 60601-1-8: 2006 + A1: 2012
IEC 60730-1: 2013
IEC 61508: 2010
IEC 62304: 2006
ISO 26262
ISO 26262: 2011
MISRA C
MISRA C: 2012 (March 2013)
xxi
Preface
This book is concerned with the development of reliable, real-time
embedded systems. The particular focus is on the engineering of systems
based on time-triggered architectures.
In the remainder of this preface, I attempt to provide answers to questions
that prospective readers may have about the book contents.
a. What is a “reliable embedded system”?
I have provided a definition of the phrase “System Fault” on Page xv.
My goal in this book is to present a model-based process for the
development of embedded applications that can be used to provide evidence
that the system concerned will be able to detect such faults and then handle
them in an appropriate manner, thereby avoiding Uncontrolled System
Failures.
The end result is what I mean by a reliable embedded system.
b. Who needs reliable embedded systems?
Techniques for the development of reliable embedded systems are – clearly
– of great concern in safety-critical markets (e.g. the automotive, medical,
rail and aerospace industries), where System Failures can have immediate,
fatal, consequences.
The growing challenge of developing complicated embedded systems in
traditional “safety” markets has been recognised, a fact that is reflected in
the emergence in recent years of new (or updated) international standards
and guidelines, including IEC 61508, ISO 26262 and DO-178C.
As products incorporating embedded processors continue to become ever
more ubiquitous, safety concerns now have a great impact on developers
working on devices that would not – at one time – have been thought to
require a very formal design, implementation and test process. As a
consequence, even development teams working on apparently “simple”
household appliances now need to address safety concerns. For example,
manufacturers need to ensure that the door of a washing machine cannot be
opened by a child during a “spin” cycle, and must do all they can to avoid
the risk of fires in “always on” applications, such as fridges and freezers.
Again, recent standards have emerged in these sectors (such as IEC 607301).
Reliability is – of course – not all about safety (in any sector). Subject to
inevitable cost constraints, most manufacturers wish to maximise the
xxiii
reliability of the products that they produce, in order to reduce the cost of
warranty repairs, minimise product recalls and ensure repeat orders. As
systems grow more complicated, ensuring the reliability of embedded
systems can present significant challenges for any organisation.
I have found that the techniques presented in this book can help developers
(and development teams) in many sectors to produce reliable and secure
systems.
c. Why work with “time-triggered” systems?
As noted at the start of this Preface, the focus of this book is on TT systems.
Implementation of a TT system will typically start with a single interrupt
that is linked to the periodic overflow of a timer. This interrupt may drive a
task scheduler (a simple form of “operating system”). The scheduler will –
in turn – release the system tasks at predetermined points in time.
TT can be viewed as a subset of a more general event-triggered (ET)
architecture. Implementation of a system with an ET architecture will
typically involve use of multiple interrupts, each associated with specific
periodic events (such as timer overflows) or aperiodic events (such as the
arrival of messages over a communication bus at unknown points in time).
TT approaches provide an effective foundation for reliable real-time systems
because – during development and after construction – it is (compared with
equivalent ET designs) easy to model the system and, thereby, determine
whether all of the key timing requirements have been met. This can help to
reduce testing costs – and reduce business risks.
The deterministic behaviour of TT systems also offers very significant
advantages at run time, because – since we know precisely what the system
should be doing at a given point in time – we can very quickly determine
whether it is doing something wrong.
xxiv
d. How does this book relate to international safety standards?
Throughout this book it is assumed that some readers will be developing
embedded systems in compliance with one or more international standards.
The standards discussed during this book include following:
п‚·
п‚·
п‚·
п‚·
п‚·
IEC 61508 (industrial systems / generic standard)
ISO 26262 (automotive systems)
IEC 60730 (household goods)
IEC 62304 (medical systems)
DO-178C (aircraft)
No prior knowledge of any of these standards is required in order to read
this book.
Please note that full references to these standards are given on p.xxi.
e. What programming language is used?
The software in this book is implemented almost entirely in �C’.
For developers using C, the “MISRA C” guidelines are widely employed as
a “language subset”, with associated coding guidelines (MISRA, 2012).
f. Is the source code “freeware”?
This book is supported by a complete set of “Time-Triggered Reference
Designs” (TTRDs).
Both the TTRDs and this book describe patented2 technology and are subject
to copyright and other restrictions.
The TTRDs provided with this book may be used without charge:
[i] by universities and colleges in courses for which a degree up to and
including “MSc” level (or equivalent) is awarded; [ii] for non-commercial
projects carried out by individuals and hobbyists.
All other use of any of the TTRDs associated with this book requires
purchase (and maintenance) of a low-cost, royalty free ReliabiliTTy
Technology Licence:
http://www.safetty.net/products/reliabilitty
2
Patents applied for.
xxv
g. How does this book relate to other books in the “ERES” series?
The focus throughout all of the books in the ERES series is on singleprogram, real-time systems.
Typical applications for the techniques described in this series include
control systems for aircraft, steer-by-wire systems for passenger cars, patient
monitoring devices in a hospital environment, electronic door locks on
railway carriages, and controllers for domestic “white goods”.
Such systems are currently implemented using a wide range of different
hardware “targets”, including various different microcontroller families
(some with a single core, some with multiple independent cores, some with
“lockstep” architectures) and various FPGA / CPLD platforms (with or
without a “soft” or “hard” processor core).
Given the significant differences between the various platforms available
and the fact that most individual developers (and many organisations) tend
to work in a specific sector, using a limited range of hardware, I decided that
I would simply “muddy the water” by trying to cover numerous
microcontroller families in a single version of this book. Instead, ERES will
be released in a number of distinct editions, each with a focus on a particular
(or small number) of hardware targets and related application sectors.
You’ll find up-to-date information about the complete book series here:
http://www.safetty.net/publications/the-engineering-of-reliable-embedded-systems
h. What processor hardware is used in this book?
In this edition of the book, the main processor target is an NXPВ® LPC1769
microcontroller. In almost all cases, the code examples can be executed on
a low-cost and readily-available evaluation platform.3
The LPC1769 is an ARMВ® Cortex-M3 based microcontroller (MCU) that
operates at CPU frequencies of up to 120 MHz.
The LPC1769 is intended for use in applications such as: industrial
networking; motor control; white goods; eMetering; alarm systems; and
lighting control.
The two case studies in this book focus on the use of the LPC1769
microcontroller in white goods (specifically, a washing machine). However,
the TT software architecture that is employed in these examples is generic in
nature and can be employed in many different systems (in various sectors).
Many of the examples employ key LPC1769 components – such as the
Memory Protection Unit – in order to improve system reliability and safety.
3
Further information about this hardware platform is presented in Appendix 2.
xxvi
i. How does this book relate to “PTTES”?
This book is not intended as an introductory text: it is assumed that readers
already have experience developing embedded systems, and that they have
some understanding of the concept of time-triggered systems. My previous
book “Patterns for Time-Triggered Embedded Systems” (PTTES) can be
used to provide background reading.4
It is perhaps worth noting that I completed work on PTTES around 15 years
ago. Since then, I estimate that I’ve worked on or advised on more than 200
�TT’ projects, and helped around 50 companies to make use of a TT
approach for the first time. I’ve learned a great deal during this process.
In the present book, I’ve done my best to encapsulate my experience (to
date) in the development of reliable, real-time embedded systems.
j. Is there anyone that you’d like to thank?
As with my previous books, I’d like to use this platform to say a public
“thank you” to a number of people.
In total, I spent 21 years in the Engineering Department at the University of
Leicester (UoL) before leaving to set up SafeTTy Systems. I’d like to thank
the following people for their friendship and support over the years:
Fernando Schlindwein, John Fothergill, Len Dissado, Maureen Strange,
Barrie Jones, Ian Postlethwaite, Andrew Norman, Simon Hogg, Simon Gill,
John Beynon, Hans Bleis, Pete Barwell, Chris Marlow, Chris Edwards,
Julie Hage, Matt Turner, Bill Manners, Paul Lefley, Alan Stocker,
Barry Chester, Michelle Pryce, Tony Forryan, Tom Robotham, Geoff
Folkard, Declan Bates, Tim Pearce, Will Peasgood, Ian Jarvis, Dan Walker,
Hong Dong, Sarah Hainsworth, Paul Gostelow, Sarah Spurgeon, Andy
Truman, Alan Wale, Alan Cocks, Lesley Dexter, Dave Siddle, Guido
Herrmann, Andy Chorley, Surjit Kaur, Julie Clayton, Andy Willby,
Dave Dryden and Phil Brown.
In the Embedded Systems Laboratory (at the University of Leicester), I had
the opportunity to work with an exceptional research team. I’d particularly
like to thank Devaraj Ayavoo, Keith Athaide, Zemian Hughes, Pete Vidler,
Farah Lakhani,
Aley Imran Rizvi,
Susan Kurian,
Musharraf Hanif,
Kam Chan, Ioannis Kyriakopoulos, Michael Short and Imran Sheikh, many
of whom I worked with for many years (both in the ESL and at TTE
Systems).
I also enjoyed having the opportunity to work with
Tanya Vladimirova, Royan Ong, Teera Phatrapornnant, Chisanga Mwelwa,
Ayman Gendy, Huiyan Wang, Muhammad Amir, Adi Maaita, Tim
4
“PTTES” can be downloaded here: http://www.safetty.net/publications/pttes
xxvii
Edwards, Ricardo Bautista-Quintero, Douglas Mearns, Yuhua Li, Noor
Azurati Ahmad, Mouaaz Nahas, Chinmay Parikh, Kien Seng Wong,
David Sewell, Jianzhong Fang and Qiang Huang.
In 2005, I was asked by staff in what became the “Enterprise and Business
Development Office” (at the University of Leicester) to begin the process
that led to the formation of TTE Systems Ltd. Tim Maskell was there from
the start, and it was always a great pleasure working with him. I also
enjoyed working with David Ward, Bill Brammar and James Hunt.
The “TTE” team involved a number of my former research colleagues, and I
also had the pleasure of working with Muhammad Waqas Raza, Anjali Das,
Adam Rizal Azwar, Rishi Balasingham, Irfan Mir, Rajas More and
Vasudevan Pillaiand Balu. At this time, I also enjoyed having the
opportunity to work with my first team of Board members and investors: I’d
particularly like to thank Alan Lamb, Clive Smith, Penny Attridge,
Jonathan Gee, Tim Maskell (again), Viv Hallam, Chris Jones and
Ederyn Williams for their support over the lifetime of the company.
Since the start of 2014, I’ve been focused on getting SafeTTy Systems off
the ground. Steve Thompson and Farah Lakhani joined me at the start of
this new project and it has been a pleasure to have the opportunity to work
with them again.
I’m grateful to Cass and Kynall (for being there when I have needed them –
I hope to return the favour before too long), and to Bruce and Biggles (for
keeping my weight down). I’d like to thank David Bowie for “The Next
Day”, Thom Yorke and Radiohead for “Kid A”, and Sigur Rós for “()”.
Last but not least, I’d like to thank Sarah for having faith in me in the last
two years, as I took our lives “off piste”.
Michael J. Pont
November 2014
xxviii
PART ONE: INTRODUCTION
“If you want more effective programmers, you will discover
that they should not waste their time debugging, they should
not introduce the bugs to start with.”
Edsger W. Dijkstra, 1972.
1
CHAPTER 1: Introduction
In this chapter we provide an overview of the material that is covered in
detail in the remainder of this book.
Design &
implement
Requirements
(functional)
Requirements
(temporal)
Test & verify
Impact
of Faults
& Hazards
Real-time
embedded
system
Run-time
monitoring
Figure 1: The engineering of reliable real-time embedded systems (overview). In this book, our
focus will be on the stages shown on the right of the figure (grey boxes).
1.1. Introduction
The process of engineering reliable real-time embedded systems (the focus
of this book) is summarised schematically in Figure 1. Projects will
typically begin by recording both the functional requirements and the timing
requirements for the system, and by considering potential faults and hazards.
Design and implementation processes will then follow, during and after
which test and verification activities will be carried out (in order to confirm
that the requirements have been met in full). Run-time monitoring will then
be performed as the system operates.
The particular focus of this book is on the development of this type of
system using time-triggered (TT) architectures.
What distinguishes TT approaches is that it is possible to model the system
behaviour precisely and – therefore – determine whether all of the timing
requirements have been met. It is important to appreciate that we can use
our models to confirm that the system behaviour is correct both during
development and at run time. This can provide a very high level of
confidence that the system will either: [i] operate precisely as required; or
[ii] move into a pre-determined “limp home” or “fail silent” mode.
In this chapter, we explain what a time-triggered architecture is, and we
consider some of the processes involved in developing such systems: these
processes will then be explored in detail in the remainder of the text.
3
1.2. Single-program, real-time embedded systems
An embedded computer system (“embedded system”) is usually based on
one or more processors (for example, microcontrollers or microprocessors),
and some software that will execute on such processor(s).
Embedded systems are widely used in a variety of applications ranging from
brake controllers in passenger vehicles to multi-function mobile telephones.
The focus in this text is on what are sometimes called “single-program”
embedded systems. Such applications are represented by systems such as
engine controllers for aircraft, steer-by-wire systems for passenger cars,
patient monitoring devices in a hospital environment, automated door locks
on railway carriages, and controllers for domestic washing machines.
The above systems have the label “single-program” because the general user
is not able to change the software on the system (in the way that programs
are installed on a laptop, or “apps” are added to a smartphone): instead, any
changes to the software in the steering system – for example – will be
performed as part of a service operation, by suitably-qualified individuals.
What also distinguishes the systems above (and those discussed throughout
this book) is that they have real-time characteristics.
Consider, for example, the greatly simplified aircraft autopilot application
illustrated schematically in Figure 2. Here we assume that the pilot has
entered the required course heading, and that the system must make regular
and frequent changes to the rudder, elevator, aileron and engine settings (for
example) in order to keep the aircraft following this path.
An important characteristic of this system is the need to process inputs and
generate outputs at pre-determined time intervals, on a time scale measured
in milliseconds. In this case, even a slight delay in making changes to the
rudder setting (for example) may cause the plane to oscillate very
unpleasantly or, in extreme circumstances, even to crash.
In order to be able to justify the use of the aircraft system in practice (and to
have the autopilot system certified), it is not enough simply to ensure that
the processing is �as fast as we can make it’: in this situation, as in many
other real-time applications, the key characteristic is deterministic
processing. What this means is that in many real-time systems we need to
be able to guarantee that a particular activity will always be completed
within – say – 2 ms (+/- 5 µs), or at 6 ms intervals (+/- 1 µs): if the
processing does not match this specification, then the application is not just
slower than we would like, it is simply not fit for purpose.
4
Reminder
0
1 second (s)
1 millisecond (ms)
1 micosecond (Вµs)
1 nanosecond (ns)
= 1.0 second (10 seconds) = 1000 ms.
-3
= 0.001 seconds (10 seconds) = 1000 Вµs.
-6
= 0.000001 seconds (10 seconds) = 1000 ns.
-9
= 0.000000001 seconds (10 seconds).
Box 1
x, y, z = position coordinates
пЃµпЂ¬пЂ пЃўпЂ¬пЂ пЃ¶пЂ = velocity cordinates
p = roll rate
q = pitch rate
r = yaw rate
q
y,пЃў
Rudder
пЃ¤r
Elevator
пЃ¤e
Aileron
пЃ¤a
x,пЃµ
p
z,пЃ¶
r
Yaw (rate)
sensor
Pitch
(rate)
sensor
Rudder
Roll
(rate)
sensor
Elevator
Aircraft
Autopilot
System
(schematic)
Main
pilot
controls
Aileron
Position
sensors
(GPS)
Main engine
(fuel)
controllers
Velocity
sensors
(3 axes)
Figure 2: A high-level schematic view of an autopilot system.
5
Tom De Marco has provided a graphic description of this form of hard realtime requirement in practice, quoting the words of a manager on a software
project:
“We build systems that reside in a small telemetry computer,
equipped with all kinds of sensors to measure electromagnetic fields
and changes in temperature, sound and physical disturbance. We
analyze these signals and transmit the results back to a remote
computer over a wide-band channel. Our computer is at one end of a
one-meter long bar and at the other end is a nuclear device. We drop
them together down a big hole in the ground and when the device
detonates, our computer collects data on the leading edge of the blast.
The first two-and-a-quarter milliseconds after detonation are the most
interesting. Of course, long before millisecond three, things have
gone down hill badly for our little computer. We think of that as a
real-time constraint.”
[De Marco, writing in the foreword to Hatley and Pirbhai, 1987]
In this case, it is clear that this real-time system must complete its recording
on time: it has no opportunity for a “second try”. This is an extreme
example of what is sometimes referred to as a �hard’ real-time system.
1.3. TT vs. ET architectures
When creating a single-program design, developers must choose an
appropriate system architecture. One such architecture is a “time-triggered”
(TT) architecture. Implementation of a TT architecture will typically
involve use of a single interrupt that is linked to the periodic overflow of a
timer. This interrupt will be used to drive a task scheduler (a simple form of
“operating system”). The scheduler will – in turn – release the system tasks
at predetermined points in time.
TT architectures can be viewed as a “safer subset” of a more general eventtriggered architecture (see Figure 3 and Figure 4). Implementation of a
system with an event-triggered architecture will typically involve use of
multiple interrupts, each associated with specific periodic events (such as
timer overflows) and aperiodic events (such as the arrival of messages over
a communication bus at random points in time). ET designs are traditionally
associated with the use of what is known as a real-time operating system (or
RTOS), though use of such a software platform is not a defining
characteristic of an ET architecture.
6
MISRA C
A “safe subset”
The C Programming Language
Figure 3: Safer language subsets (for example, MISRA C) are employed by many organisations in
order to improve system reliability. See MISRA (2012).
Time-Triggered Systems
A “safe subset”
Event-Triggered Systems
Figure 4: In a manner similar to MISRA C (Figure 3), TT approaches provide a “safer subset” of ET
designs, at the system architecture level.
Whether TT or ET architectures are employed, the system tasks are typically
named blocks of program code that perform a particular activity (for
example, a task may check to see if a switch has been pressed): tasks are
often implemented as functions in programming languages such as �C’ (and
this is the approached followed in the present book).
It should be noted that – at the time of writing (2014) – the use of ET
architectures and RTOS solutions is significantly more common than the use
of TT solutions, at least in projects that are not safety related.
1.4. Modelling system timing characteristics
TT computer systems execute tasks according to a predetermined task
schedule. As noted in Section 1.3, TT systems are typically (but not
necessarily) implemented using a design based on a single interrupt linked
to the periodic overflow of a timer.
For example, Figure 5 shows a set of tasks (in this case Task A, Task B,
Task C and Task D) that might be executed by a TT computer system
according to a predetermined task schedule.
7
A
C
B
D
A
C
B
D
...
Time
Figure 5: A set of tasks being released according to a pre-determined schedule.
In Figure 5, the release of each sub-group of tasks (for example, Task A and
Task B) is triggered by what is usually called a timer “tick”. In most
designs (including all of those discussed in detail in this book), the timer
tick is implemented by means of a timer interrupt. These timer ticks are
periodic. In an aerospace application, the “tick interval” (that is, the time
interval between timer ticks) of 25 ms might be used, but shorter tick
intervals (e.g. 1 ms or 100 Вµs) are more common in other systems.
In Figure 5, the task sequence executed by the computer system is as
follows: Task A, Task C, Task B, Task D. In many designs, such a task
sequence will be determined at design time (to meet the system
requirements) and will be repeated “forever” when the system runs (until the
system is halted or powered down, or a System Failure occurs).
Sometimes it is helpful (not least during the design process) to think of this
task sequence as a “Tick List”: such a list lays out which task will run in
each system “Tick”.
For example, the Tick List corresponding to the task set shown in Figure 5
could be represented as follows:
[Tick 0]
Task A
Task C
[Tick 1]
Task B
Task D
Once the system reaches the end of the Tick List, it starts again at the
beginning.
In Figure 5, the tasks are co-operative (or “non-preemptive”) in nature: each
task must complete before another task can execute. The design shown in
these figures can be described as “time triggered co-operative” (TTC) in
nature.
We say more about designs that involve task pre-emption in Section 1.6.
8
The importance of Tick Lists
The creation and use of Tick Lists is central to the engineering of reliable TT
systems.
Through the use of this simple model, we can determine key system characteristics
– such as response times, task jitter levels and maximum CPU loading – very early
in the design process.
We can then continue to check these characteristics throughout the development
process, and during run-time operation of the system.
We will consider the creation and use of Tick Lists in detail in Chapter 4.
Box 2
int main(void)
{
SYSTEM_Init();
1 ms timer
SCH_Start();
while(1)
{
SCH_Dispatch_Tasks();
}
void SCH_Update(void)
{
Tick_count++;
}
return 1;
}
Figure 6: A schematic representation of the key components in a TTC scheduler.
1.5. Working with “TTC” schedulers
Many (but by no means all) TT designs are implemented using co-operative
tasks and a “TTC” scheduler.
Figure 6 shows a schematic representation of the key components in such a
scheduler. First, there is function SCH_Update(): in this example, this is
linked to a timer that is assumed to generate periodic “ticks” – that is, timer
interrupts – every millisecond.
The SCH_Update() function is responsible for keeping track of elapsed
time.
Within the function main() we assume that there are functions to initialise
the scheduler, initialise the tasks and then add the tasks to the schedule.
In Figure 6, function main(), the process of releasing the system tasks is
carried out in the function SCH_Dispatch_Tasks().
The operation of a typical SCH_Dispatch() function is illustrated
schematically in Figure 7. In this figure, the dispatcher begins by
determining whether there is any task that is currently due to run. If the
answer to this question is “yes”, the dispatcher runs the task.
9
Start task
dispatcher
Determine which
task is due to run
next in this tick
(if any)
Run task
Is there a
task due to
run?
Yes
No
Go to sleep
End task dispatcher
Figure 7: The operation of a task dispatcher.
The dispatcher repeats this process until there are no tasks remaining that are
due to run. The dispatcher then puts the processor to sleep: that is, it places
the processor into a power-saving mode. The processor will remain in this
power-saving mode until awakened by the next timer interrupt: at this point
the timer ISR (Figure 6) will be called again, followed by the next call to the
dispatcher function (Figure 7).
It should be noted that there is a deliberate split between the process of timer
updates and the process of task dispatching (shown in main() in Figure 6 and
described in more detail in Figure 7). This split means that it is possible for
the scheduler to execute tasks that are longer than one tick interval without
missing timer ticks. This gives greater flexibility in the system design, by
allowing use of a short tick interval (which can make the system more
responsive) and longer tasks (which can simplify the design process). This
split may also help to make the system more robust in the event of run-time
errors: we say more about this in Chapter 2.
Flexibility in the design process and the ability to recover from transient
errors are two reasons why “dynamic” TT designs (with a separate timer
ISR and task dispatch functions) are generally preferred over simpler
designs in which tasks are dispatched from the timer ISR.
P
Tick 0
A
P
Tick 1
B
P
Tick 2
A
P
B
Tick 3
Figure 8: Executing tasks using a TTH scheduler. See text for details.
10
...
Time
1.6. Supporting task pre-emption
The designs discussed in Section 1.4 and Section 1.5 involve co-operative
tasks: this means that each task “runs to completion” after it has been
released. In many TT designs, high-priority tasks can interrupt (pre-empt)
lower-priority tasks.
For example, Figure 8 shows a set of three tasks: Task A (a low-priority, cooperative task), Task B (another low-priority, co-operative task), and Task P
(a higher-priority pre-empting task). In this example, the lower-priority
tasks may be pre-empted periodically by the higher-priority task. More
generally, this kind of “time triggered hybrid” (TTH) design may involve
multiple co-operative tasks (all with an equal priority) and one or more preempting tasks (of higher priority).
We can also create “time-triggered pre-emptive” (TTP) schedulers: these
support multiple levels of task priority.
We can – of course – record the Tick List for TTH and TTP designs. For
example, the task sequence for Figure 8 could be listed as follows: Task P,
Task A, Task P, Task B, Task P, Task A, Task P, Task B.
We say more about task pre-emption in Part Three.
1.7. Different system modes
Almost all practical embedded systems have at least two system modes,
called something like “Normal mode” and “Fault mode”. However most
have additional system modes. For example, Figure 9 shows a schematic
representation of the software architecture for an aircraft system with system
modes corresponding to the different flight stages (preparing for takeoff,
climbing to cruising height, etc).
In this book, we consider that the system mode has changed if the task set
has changed. It should therefore be clear that we are likely to have a
different Tick List for each system mode.
Please note that – even in a TT design – the timing of the transition between
system modes is not generally known in advance (because, for example, the
time taken for the plane shown in Figure 9 to reach cruising height will vary
with weather conditions), but this does not alter the development process.
The key feature of all TT designs is that – whatever the mode – the tasks are
always released according to a schedule that is determined, validated and
verified when the system is designed.
11
Figure 9: An example of a system with multiple operating modes.
We say more about system modes in Chapter 7.
1.8. A “Model-Build-Monitor” methodology
A three-stage development process is described in detail during the course
of this book.
The first stage involves modelling the system (using one or more Tick Lists,
depending on the number of modes), as outlined in Section 1.4. The second
stage involves building the system (for example, using a simple TTC
scheduler, as outlined in Section 1.5). The third stage involves adding
support for run-time monitoring.
Run-time monitoring is essential because we need to ensure that the
computer system functions correctly in the event that Hardware Faults occur
(as a result, for example, of electromagnetic interference, or physical
damage: see “Definitions” on Page xv). In addition, as designs become
larger, it becomes unrealistic to assume that there will not be residual
Software Errors in products that have been released into the field: it is
clearly important that there should not be an Uncontrolled System Failure in
the event that such errors are present. Beyond issues with possible
Hardware Faults and residual errors in complex software, we may also need
to be concerned about the possibility that attempts could be made to
introduce Deliberate Software Changes into the system, by means of
“computer viruses” and similar attacks.
The approach to run-time monitoring discussed in this book involves
checking for resource-related faults and / or time-related faults (Figure 10).
As an example of resource-related fault, assume that Pin 1-23 on our
microcontroller is intended to be used exclusively by Task 45 to activate the
steering-column lock in a passenger vehicle. This lock is intended to be
engaged (to secure the vehicle against theft) only after the driver has parked
and left the vehicle.
12
Resources
Resource Barrier
(Protected against
resource
corruption)
Resource Barrier
+
Time Barrier
(Fully protected)
Time Barrier
No protection
(Protected against
task overruns,
underruns, and
sequence errors)
Time
Figure 10: Run-time monitoring is employed to guard against Uncontrolled System Failures.
In this book, we will check for faults that are evidenced by incorrect CPU usage (task overruns,
task underruns and task sequencing) as well as faults related to other hardware resources
(such as CAN controllers, analogue-to-digital convertors, port pins and memory).
A (potentially very serious) resource-related fault would occur if Pin 1-23
was to be activated by another task in the system while the vehicle was
moving at high speed.
Of course, both ET and TT designs need to employ mechanisms to check for
resource-related faults. However, this process is much more easily modelled
(and therefore more deterministic) in TT designs. For instance, in “TTC”
designs, precise control of access to shared resources is intrinsic to the
architecture (because all tasks “run to completion”). Even in TT designs
that involve task pre-emption, controlling access to shared resources is –
generally – more straightforward than in ET designs. One very important
consequence is that while the impact of priority inversion (PI) can be
ameliorated in ET designs through the use of mechanisms such as “ceiling
protocols” (as we will discuss in Chapter 12), PI problems can be eliminated
only through the use of a TT solution.
Beyond this, we have found that a design approach based on the concept of
“Task Contracts” can help developers to implement effective “resource
barriers” in TT systems. We say more about this approach in Chapter 8.
In addition to resource-related faults, we also need to consider timing related
faults (please refer again to Figure 10). Here, a second – very significant –
advantage of TT designs comes into play.
As we have seen in this chapter, TT systems are – by definition – designed
to execute sets of tasks according to one or more pre-determined schedules:
in each system mode, the required sequence of tasks is known in advance.
During the design process, the task schedules are carefully reviewed and
assessed against the system requirements, and at run time a simple task
scheduler is used to release the tasks at the correct times. If the task set is
not then executed in the correct sequence at run time, this may be
symptomatic of a serious fault. If such a situation is not detected quickly,
13
this may have severe consequences for users of the system, or those in the
vicinity of the system.
For example, consider that we have detected a fault in the braking system of
a passenger car: if the driver is already applying the brakes in an emergency
situation when we detect the fault, the fault-detection mechanism is of little
value. Similarly, late detection (or lack of detection) of faults in aerospace
systems, industrial systems, defence systems, medical systems, financial
systems or even household goods may also result in injury, loss of human
life and / or very significant financial losses.
Using the techniques presented in Chapter 10, we can perform “predictive
monitoring” of the task execution sequence during the system operation. In
many cases, this means that we can detect that the system is about to run an
incorrect task before this task even begins executing.
This type of solution can greatly simplify the process of achieving
compliance with international standards and guidelines. For example, to
achieve compliance with the influential IEC 61508 standard, many designs
require the use of a “diverse monitor” unit. Such a unit is intended to
prevent the system from entering an unsafe state5, which is precisely what
we can achieve using predictive monitoring of a TT architecture.
Similar requirements arise from other standards (for example, the need to
implement “Safety Mechanisms” in ISO 26262).
1.9. How can we avoid Uncontrolled System Failures?
As we conclude this introductory chapter, we’ll consider one of the key
challenges facing developers of modern embedded systems.
As we have discussed in previous sections, embedded systems typically
consist of: [i] a set of tasks; [ii] a scheduler, operating system or similar
software framework that will have some control over the release of the
tasks.
In an ideal world, the resulting architecture might look something like that
illustrated in Figure 11 (left). From the developer’s perspective, such a
design may be attractive, because each software component is isolated: this
– for example – makes run-time monitoring straightforward, and means that
it is easy to add further tasks to the system (for example, during
development or later system upgrades) with minimal impact on the existing
tasks.
5
It is sometimes assumed that a “diverse monitor” is intended to detect when a system has
entered into an unsafe state, but that is not what is required in IEC 61508 [2010].
14
Task A
Task A
Task B
Scheduler
Scheduler
Task B
Figure 11: An “ideal” implementation of an embedded system with two tasks (left)
along with a more practical implementation (right).
In practice, such a “one processor per task” design would prove to be
impossibly expensive in most sectors. Instead, we are likely to have
multiple tasks and the scheduler / operating system sharing the same
processor (Figure 11, right).
In this real-world design, a key design aim will be to ensure that no task can
interfere with any other task, and that no task can interfere with the
scheduler (and vice versa): this is sometimes described as a goal of
“Freedom From Interference” (FFI).
FFI is a very worthy goal. Unfortunately, in any non-trivial embedded
system, there are a great many ways in which it is possible for tasks (and
other software components) that share a CPU, memory and other resources
to interact. As a consequence, any claim that we can prevent any
interference would be likely to be met with some scepticism.
This does not mean that we need to dismiss FFI as “unachievable”, because
– while interference may not be preventable – it may be detectable.
More specifically, we will argue in this book that – through use of an
appropriate implementation of a TT design, with a matched monitoring
system – we will often be able to meet FFI requirements. We can do this
because the engineering approach described in the following chapters can be
used to: [i] provide evidence of the circumstances in which we will be able
to detect any interference between tasks, or between tasks and the scheduler,
in a given system; and [ii] provide evidence of our ability to move the
system into pre-determined “limp home” or “fail silent” operating mode in
the event that such interference is detected.
Overall, the goal – in this FFI example and with all of the systems that we
consider in this book – is to prevent Uncontrolled System Failures, by
building upon the system architecture illustrated schematically in Figure 12.
15
Task
AAA
Task
Task
Task
AAA
Task
Task
Task
AAA
Task
Task
Task
Task
TaskAAA
TT Monitoring
System
TT Scheduler
Figure 12: A schematic representation of the architecture that we will use in this book in order to
avoid Uncontrolled System Failures.
1.10. Conclusions
In this introductory chapter, we’ve provided an overview of the material that
is covered in detail in the remainder of this book.
In Chapter 2, we will introduce our first task scheduler.
16
CHAPTER 2: Creating a simple TTC scheduler
In Chapter 1, we noted that the implementation of most time-triggered
embedded systems involves the use of a task scheduler. In this chapter, we
explore the design of a first simple scheduler for use with sets of periodic,
co-operative tasks.
Related TTRDs
A list of the TTRDs discussed in this chapter is included below.
C programming language (LPC1769 target):
 TTRD02a: TTC scheduler with WDT support and �Heartbeat’ error reporting
п‚· TTRD02b: TTC scheduler with injected task overrun
п‚· TTRD02c: TTC scheduler (porting example)
2.1. Introduction
In this chapter, we’ll start by exploring the TTC scheduler “TTRD02a”. To
introduce TTRD02a, we will present a simple “Heartbeat” example in which
the scheduler is used to flash an LED with a 50% duty cycle and a flash rate
of 0.5 Hz: that is, the LED will be “on” for 1 second, then “off” for one
second, then “on” for one second, and so on (Figure 13).
Figure 14 provides an overview of the structure and use of this scheduler.
If you have previously used one of the schedulers described in “Patterns for
Time-Triggered Embedded Systems” (Pont, 2001), then much of the
material presented in this chapter will be familiar. However, there are some
important differences between the PTTES schedulers and those presented in
this chapter. The main differences arise as a result of the new system
foundation: this is illustrated schematically in Figure 13.
Figure 13: A schematic representation of a microcontroller running a TTC scheduler and executing
a “Heartbeat” task.
17
int main(void)
{
SYSTEM_Init();
SysTick Timer
SCH_Start();
while(1)
{
SCH_Dispatch_Tasks();
}
void SysTick_Handler(void)
{
Tick_count++;
}
return 1;
}
Figure 14: An overview of the structure and use of a TTC scheduler.
Processor
Reset
Mechanism
System-Mode Data
Processing
Unit
Processor
Configuration
Mechanism
Figure 15: The foundation of the scheduler designs presented in this book.
Perhaps the most immediately obvious differences between the PTTES
schedulers and those described here can be seen in the way that the
SCH_Update() function and SCH_Dispatch() function are structured: please
refer to Section 2.5 and Section 2.7, respectively, for further details.
There is also a significant difference in the recommended initialisation
process for the schedulers presented in this book when compared with the
approach presented in PTTES. More specifically, we now recommend that
the system initialisation process is split between function main() and
function SYSTEM_Init().
An example of a suitable implementation of function main() is shown in
Figure 14.
We will consider the scheduling functions that are called in main –
SCH_Start() and SCH_Dispatch_Tasks() – shortly. In this section, we will
focus on the system initialisation function (see Listing 7, on p.48)6.
The first point to note about the initialisation process illustrated in this
chapter is that the system has two operating modes: “FAIL_SILENT” and
“NORMAL”. In this example, any reset that is caused by the watchdog
timer (WDT) causes the system to enter the FAIL_SILENT mode, while a
normal power-on reset (and any other reset events) cause the system to enter
NORMAL mode.
6
Throughout this book, longer code listings are located at the end of each chapter.
18
In FAIL_SILENT mode, the system simply “halts” (Code Fragment 1).7
case FAIL_SILENT:
{
// Reset caused by WDT
// Trigger "fail silent" behaviour
SYSTEM_Perform_Safe_Shutdown();
Code Fragment 1: Entering “Fail_Silent” mode.
There really isn’t very much more that we can do in this mode in the
Heartbeat demo, but – in a real system design – this should be where we end
up if a serious error has been detected by the system (and no other way of
handling this error has been identified). Deciding what to do in these
circumstances requires careful consideration during the system development
process.8
As a starting point, we need to consider what to do with the system port pins
in these circumstances. Our general goal is to ensure that the pins are left in
a state where they can do minimum damage. For example, it may be
possible to turn off any dangerous equipment that is under the control of the
computer system by setting appropriate levels on the port pins.
Other options may also need to be considered. For example, if the computer
system is connected to other devices or equipment over a computer network
(wired or wireless) we may wish to try and send out a message to the other
components to indicate that the computer system has failed.
When the system reset is not caused by the WDT then – in this example –
we enter NORMAL mode.9
In this mode, we need to do the following to initialise the system:
п‚·
п‚·
п‚·
п‚·
set up the WDT, and associated WATCHDOG_Update() task;
set up the scheduler;
call the initialisation functions for all other tasks; and,
add the tasks to the schedule.
In our example, we first set up the watchdog timer.
7
8
9
You will find the code for the function SYSTEM_Perform_Safe_Shutdown() in Listing 7
(on p.46).
We will consider this matter in detail in Chapter 13.
In many system designs, there will be multiple operating modes. We consider how such
designs can be implemented in Chapter 7.
19
When used as described in this chapter, a WDT is intended to force a
processor reset (and – thereby – place the system into a FAIL_SILENT
mode) under the following circumstances:
п‚·
п‚·
when the system becomes overloaded, usually as a result of one or more
tasks exceeding their predicted “worst-case execution time” (WCET):
this is a task overrun situation; or,
when the system fails to release the WATCHDOG_Update() task
according to the pre-determined schedule for any other reason.
In TTC designs where it has been determined that – for every tick in the
hyperperiod – the sum of the task execution times of the tasks that are
scheduled to run is less than the tick interval, then a WDT timeout period
very slightly larger than the tick interval is often used.
Our example design comfortably meets the above execution-time criteria
and has a tick interval of 1 ms: we therefore set the watchdog timeout to just
over 1 ms, as follows:
// Set up WDT (timeout in *microseconds*)
WATCHDOG_Init(1100);
We’ll look at the details of the WDT configuration in Section 2.11.
Following configuration of this timer, we then set up the scheduler with
1 ms ticks, using the SCH_Init() function:
// Set up scheduler for 1 ms ticks (tick interval *milliseconds*)
SCH_Init(1);
Note that if the system cannot be configured with the required tick interval,
we force a system reset (using the WDT unit): following the reset, the
system will then enter a FAIL_SILENT mode. This type of WDT-induced
mode change will be common in many of the designs that we consider in
this book (in various different circumstances).
We will provide further information about the SCH_Init() function in
Section 2.4.
Assuming that initialisation of the scheduler was successful, we then prepare
for the Heartbeat task, by means of the HEARTBEAT_Init() function.
Further information is provided about the Heartbeat task in Section 2.13: for
now, we will simply assume that this is used to configure an I/O pin that has
been connected to LED2 on the LPC1769 board (please refer to Appendix 2
for details).
20
2.2. A first TTC scheduler (TTRD02a)
Having considered, in outline, how the system will be initialised, we now
consider the internal structure and operation of the scheduler itself.
The TTRD02a scheduler presented in this chapter is made up of the
following key components:
п‚·
п‚·
п‚·
п‚·
п‚·
A scheduler data structure.
An initialisation function.
An interrupt service routine (ISR), used to update the scheduler.
A function for adding tasks to the schedule.
A dispatcher function that releases tasks when they are due to run.
We consider each of the required components in the sections that follow.
SCH_MAX_TASKS
You will find SCH_MAX_TASKS in “Scheduler Header” file in all designs in this book.
This constant must be set to a value that is at least as large as the number of tasks
that are added to the schedule: this process is not automatic and must be checked
for each project.
Box 3
2.3. The scheduler data structure and task array
At the heart of TTRD02a is a user-defined data type (sTask) that collects
together the information required about each task.
Listing 4 shows the sTask implementation used in TTRD02a.
The task list is then defined in the main scheduler file as follows:
sTask SCH_tasks_G[SCH_MAX_TASKS];
The members of sTask are documented in Table 1.
2.4. The �Init’ function
The scheduler initialisation function is responsible for:
п‚·
п‚·
п‚·
initialising the global error variable;
initialising the task array; and,
configuring the scheduler tick.
The initialisation process begins as shown in Code Fragment 2.
21
// Reset the global error variable
Error_code_G = 0;
for (i = 0; i < SCH_MAX_TASKS; i++)
{
SCH_tasks_G[i].pTask = 0;
}
Code Fragment 2: The start of the scheduler initialisation process.
As in the PTTES schedulers, the global error variable (Error_code_G) is
used to report errors, usually via the Heartbeat task: please see Section 2.13
for further information about this.
Table 1: The members of the sTask data structure (as used in TTRD02a).
Member
Description
void (*pTask)(void)
A pointer to the task that is to be scheduled.
The task must be implemented as a “void void” function.
See Section 2.13 for an example.
uint32_t Delay
The time (in Ticks) before the task will next execute.
uint32_t Period
The task period (in Ticks).
uint32_t WCET
The worst-case execution time for the task (in Вµs).
Please note that this information is not used directly in
TTRD02a, but is employed during schedulability analysis
(see Chapter 4). In addition, in later schedulers in this
book, this information is used to assist in the detection of
run-time errors (see Chapter 9 and Chapter 10).
uint32_t BCET
The best-case execution time for the task (in Вµs).
Again, this information is not used directly in TTRD02a, but
is employed during schedulability analysis and – in later
schedulers – it is used to assist in the detection of run-time
errors.
uint32_t RunMe
This flag is set to 1 by the scheduler when the task is due
to run.
The error codes themselves can be found in the Project Header file (main.h,
Listing 1).
The next step in the scheduler initialisation process involves setting up the
timer Ticks.
In TTRD02a, this code is based on the ARM CMSIS10.
10
CortexВ® Microcontroller Software Interface Standard.
22
As part of this standard, ARM provides a template file system_device.c that
must be adapted by the manufacturer of the corresponding microcontroller
to match their device.
At a minimum, system_device.c must provide:
п‚·
п‚·
a device-specific system configuration function, SystemInit(); and,
a global variable that represents the system operating frequency,
SystemCoreClock.
The SystemInit() function performs basic device configuration, including
(typically) initialisation of the oscillator unit (PLL). The SystemCoreClock
value is then set to match the results of this configuration process.
In TTRD02a, we record our expected system operating frequency in main.h
by means of Required_SystemCoreClock. We then check that the system
has been configured as expected, as shown in Code Fragment 3.
As we enable the WDT unit before we call SCH_Init(), we can force a reset
(and a transition to FAIL_SILENT mode) if – for some reason – the system
operating frequency is not as expected.
CMSIS also provides us with a SysTick timer to drive the scheduler, and a
means of configuring this timer to give the required Tick rate (Code
Fragment 3).
// Now to set up SysTick timer for "ticks" at interval TICKms
if (SysTick_Config(TICKms * SystemCoreClock / 1000))
{
// Fatal error
...
while(1);
}
Code Fragment 3: Configuring the SysTick timer.
A key advantage of using the “SysTick” to drive your scheduler is that this
approach is widely used and very easily portable between microcontroller
families.
Please refer to Section 2.17 for information about the use of other timers as
a source of system ticks.
2.5. The �Update’ function
Code Fragment 4 shows the SysTick ISR.
23
void SysTick_Handler(void)
{
// Increment tick count (only)
Tick_count_G++;
}
Code Fragment 4: The �Update’ function (SysTick_Handler()) from TTRD02a.
This arrangement ensures that the scheduler function can keep track of
elapsed time, even in the event that tasks execute for longer than the tick
interval.
Note that the function name (SysTick_Handler) is used for compatibility
with CMSIS.
2.6. The �Add Task’ function
As the name suggests, the �Add Task’ function – Listing 5 - is used to add
tasks to the task array, to ensure that they are called at the required time(s).
The function parameters are (again) as detailed in Table 1.
Please note that this version of the scheduler is “less dynamic” than the
version presented in PTTES. One change is that only periodic tasks are
supported: “one shot” tasks can no longer be scheduled. This, in turn,
ensures that the system can be readily modelled (at design time) and
monitored (at run time), processes that we will consider in detail in
subsequent chapters.
We say more about the static nature of the schedulers in this book in
Section 2.10.
2.7. The �Dispatcher’
The release of the system tasks is carried out in the function
SCH_Dispatch_Tasks(): please refer back to Figure 14 to see this function
in context.
The operation of this “dispatcher” function is illustrated schematically in
Figure 16.
24
Start task
dispatcher
Determine which
task is due to run
next in this tick
(if any)
Run task
Is there a
task due to
run?
Yes
No
Go to sleep
End task dispatcher
Figure 16: The operation of the dispatcher in the TTC scheduler described in this chapter.
In Figure 16 the dispatcher begins by determining whether there is a task
that is currently due to run. If the answer to this question is “yes”, the
dispatcher runs the task. It repeats this process (check, run) until there are
no tasks remaining that are due to run.
The dispatcher then puts the processor to sleep: that is, it places the
processor into a power-saving mode. The processor will remain in this
power-saving mode until awakened by the next timer interrupt: at this point
the timer ISR will be called again, followed by the next call to the dispatcher
function (Figure 14).
The SCH_Dispatch function employed in TTRD02a is shown in Listing 5.
Referring again to Figure 14, it should be noted that there is a deliberate
split between the process of timer updates and the process of task
dispatching.
This division means that it is possible for the scheduler to execute tasks that
are longer than one tick interval without missing timer ticks. This gives
greater flexibility in the system design, by allowing use of a short tick
interval (which can make the system more responsive) and longer tasks
(which can – for example – simplify the design process).
Although this flexibility is available in the scheduler described in this
chapter, many (but not all) TTC systems are designed to ensure that no tasks
are running when a timer interrupt occurs: however, even in such designs, a
run-time errors task may mean that a task takes longer to complete. Because
25
of the dynamic nature of the scheduler, the system may be able to recover
from such run-time errors, provided that the error is not permanent.
Flexibility in the design process and the ability to recover from errors are
two reasons why “dynamic” TT designs (with a separate timer ISR and task
dispatch functions) are generally preferred over simpler designs in which
tasks are dispatched from the timer ISR.
In addition, separating the ISR and task dispatch functions also makes it
very simple to create TT designs with support for task pre-emption
(including “time-triggered hybrid” – TTH – architectures): we discuss this
process in Chapter 11.
In this listing, please note that Tick_count_G is a “shared resource”: it is
accessed both in the scheduler Update ISR and in this Dispatcher function.
To avoid possible conflicts, we disable interrupts before accessing
Tick_count_G in the Dispatcher.
2.8. The �Start’ function
The scheduler Start function (Code Fragment 5) is called after all of the
required tasks have been added to the schedule.
void SCH_Start(void)
{
// Enable SysTick timer
SysTick->CTRL |= 0x01;
// Enable SysTick interrupt
SysTick->CTRL |= 0x02;
}
Code Fragment 5: The SCH_Start() function from TTRD02a. This function should be
called after all required tasks have been added to the schedule.
SCH_Start() starts the scheduler timer, and enables the related interrupt.
2.9. The �sleep’ function
In most cases, the scheduler enters “idle” mode at the end of the Dispatcher
function: this is achieved by means of the SCH_Go_To_Sleep() function
(Code Fragment 6).
void SCH_Go_To_Sleep()
{
// Enter sleep mode = "Wait For Interrupt"
__WFI();
}
Code Fragment 6: The SCH_Go_To_Sleep() function from TTRD02a.
The system will then remain “asleep” until the next timer Tick is generated.
26
Clearly, the use of idle mode can help to reduce power consumption.
However, a more important reason for putting the processor to sleep is to
control the level of “jitter” in the Tick timing.
The central importance of jitter in the system operation will be explored in
Chapter 4. In Chapter 5, we will explore the use of idle mode in schedulers
in more detail.
2.10. Where is the “Delete Task” function?
Traditional approaches to changing system modes in TT designs involve
mechanisms for adding / removing tasks from the schedule. For example,
the TT task scheduler described in “PTTES” provides SCH_Add_Task() and
SCH_Delete_Task() functions that can be called at any time while the
scheduler is running.
Such mechanisms for changing system modes have the benefit of simplicity.
We will argue throughout this book that – in general – “A Good Thing”.
However, the author has had the opportunity to review many system designs
created using variations on the PTTES schedulers over the years: in doing
so, it has become clear that providing an easy way of changing the task set at
run-time has had unintended consequences.
TT schedules are – by their very nature – static in nature, and a key strength
of this development approach is that a complete task schedule can be
carefully reviewed at design time, in order to confirm that all system
requirements have been met: we consider this process in detail in Chapter 4.
Once the system is in the field, we can then perform monitoring operations
to ensure that the run-time behaviour is exactly as expected at design time:
we begin to consider how we can achieve this in Chapter 9.
In general, it is extremely difficult to change the system mode in TT designs
using conventional methods without significantly undermining this static
design process. When tasks can be added or removed from the schedule at
“random” times (perhaps – for example – in response to external system
events), then the system design becomes dynamic (in effect, it is no longer
“time triggered”), and it is not generally possible to predict the precise
impact that the mode change will have on all tasks in the schedule.
Even where the perturbations to the behaviour of a TT system during
traditional mode changes are short lived, this may still have significant
consequences. TT designs are often chosen for use in systems where
security is an important consideration. In such designs – because the task
schedule is known explicitly in advance – it is possible to detect even very
small changes in behaviour that may result from security breaches (for
example, if the system code has been changed as the result of a virus, etc).
27
In circumstances where dynamic changes to a task set are permitted (as in
traditional mode changes), this may mask security-related issues.
In the approach to system mode changes recommended in this book, we
always change task sets (and – therefore – the system mode) by means of a
processor reset. This ensures that the transition is made between one
complete set of tasks (that can be subject to detailed analysis, test and
verification at design time) and another complete set of tasks (that can be
similarly assessed at design time).
2.11. Watchdog timer support
As noted throughout this chapter, TTRD02a requires a watchdog timer. The
code used to initialise the LPC1769 watchdog is shown in Listing 11.
The configuration code for the watchdog used in the demo system is
straightforward, but it should be noted that this feature of the design is
controlled by a jumper: this is needed because watchdog support cannot be
enabled when the system is debugged over the JTAG link (if the watchdog
is enabled, the processor resets will keep breaking the debug connection).
To use this design, you need to insert the jumper (between the pin identified
in the Port Header file and ground) in order to enable watchdog support).
Please refer to Code Fragment 7 and Figure 17 for further information about
this.
// Add jumper wire on baseboard to control WDT
// WDT is enabled *only* if jumper is in place.
// (Jumper is read at init phase only)
// Port 2, Pin 3
#define WDT_JUMPER_PORT (2)
#define WDT_JUMPER_PIN (0b1000)
Code Fragment 7: WDT jumper settings from the Port Header file (see Listing 1).
Ground
Port 2, Pin 3
(PIO2_3)
Figure 17: Jumper connections on the EA Baseboard that can be used to enable WDT support.
Please refer to Appendix 2 for further information about the baseboard.
The code used to refresh the watchdog is shown in Code Fragment 8. Please
note that interrupts are disabled while the WDT is “fed”, to avoid the
28
possibility that the timer ISR will be called during this operation. This
might be possible (even in a TTC design) in the event of a task overrun.
void WATCHDOG_Update(void)
{
// Feed the watchdog
__disable_irq(); // Avoid possible interruption
LPC_WDT->WDFEED = 0xAA;
LPC_WDT->WDFEED = 0x55;
__enable_irq();
}
Code Fragment 8: The WATCHDOG_Update() function from TTRD02a.
2.12. Choice of watchdog timer settings
It’s clearly important to select appropriate timeout values for the watchdog
timer (WDT) and to refresh this timer in an appropriate way (at an
appropriate interval).
One way to do this is to set up a WDT refresh task (like that shown in Code
Fragment 8) and schedule this to be released at the start of every tick (Figure
18).
W
A
W
W
B
C
W
D
...
Time
Figure 18: Running a WDT refresh task (shown as Task W) at the start of each tick interval.
In this case, we would probably wish to set the WDT timeout values to
match the tick interval. In this way, a task overrun would delay the WDT
refresh and cause a transition to a FAIL_SILENT mode (via a WDT reset):
see Figure 19.
WDT reset
W
A
W
...
B
Time
Figure 19: A WDT reset caused by a task overrun.
Used in this way, the WDT provides a form of “task guardian” (TG) that can
detect and handle task overruns (that is, tasks that – at run time – exceed the
WCET figures that were predicted when the system was constructed). Such
overruns can clearly have a very significant impact on the system operation.
Please note that this form of TG implementation is effective, but is a rather
“blunt instrument”: in Chapter 9 we’ll begin to explore some techniques that
29
will allow us to identify (and, if required, replace) individual tasks that
overrun by more than a few microseconds.
Please also note that – in addition to ensuring that we make a “clean” change
between task sets – using a reset between system modes allows us to use
appropriate watchdog settings for each system mode. This is possible
because the majority of modern COTS processors allow changes to WDT
settings (only) at the time of a processor reset.
We will provide examples of designs that use different watchdog timeouts in
different modes in Chapter 7.
2.13. The �Heartbeat’ task (with error reporting)
How can you be sure that the scheduler and microcontroller in your
embedded system is operating correctly i.e. how can you get a tangible
indication of your systems “health”?
One way to do this is to hook up a JTAG connection to the board, or some
other connection (e.g. via a USB port or a UART-based link). Such
connections may be straightforward on a workbench during system
prototyping, but are not always easy once the embedded processor has been
incorporated into a larger piece of equipment.
One effective solution is to implement a “Heartbeat” LED (e.g. see Mwelwa
and Pont, 2003).
A Heartbeat LED is usually implemented by means of a simple task that
flashes an LED on and off, with a 50% duty cycle and a frequency of
0.5 Hz: that is, the LED runs continuously, on for one second, off for one
second, and so on.
Use of this simple technique ensures that the development team, the
maintenance team and, where appropriate, the users, can tell at a glance that
the system has power, and that the scheduler is operating normally.
In addition, during development, there are two less significant (but still
useful) side benefits:
п‚·
п‚·
After a little practice, the developer can often tell “intuitively” – by
watching the LED – whether the scheduler is running at the correct rate:
if it is not, it may be that the timers have not been initialised correctly,
or that an incorrect crystal frequency has been assumed.
By adding the Heartbeat LED task to the scheduler array after all other
tasks have been included. This allows the developer to easily see that
the task array is adequately large for the needs of the application (if the
array is not large enough, the LED will never flash).
30
We can take this approach one step further, by integrating the Heartbeat task
with an error-reporting function (see Listing 9). To do this, we maintain a
(global) error variable in the system, and set this to a non-zero value in the
event of an error. We then change the LED display from the normal
Heartbeat output to an error display, in the event of an error. For example,
we could display “Error Code 2” on the LED as shown in Figure 20.
Figure 20: Output from an Error LED (displaying “Error Code 2”).
2.14. Detecting system overloads (TTRD02b)
When we are using a TTC scheduler, we will generally aim to ensure that all
tasks that are scheduled to execute in a given tick have completed their
execution time by the end of the tick.
For example, consider Figure 5. In the third tick interval, we would
generally expect that the sum of the worst-case execution time of Task E
and Task F would be less than the tick interval.
Total (maximum) task execution time
A
B
C
D
E
F
G
Tick interval
H
...
Time
Figure 21: In most TTC designs, we expect that all tasks released in a given tick will complete their
execution by the end of the tick.
It is very easy to check this during the system execution.
To do so, we set a flag every time a task is released and clear the flag when
the task completes. We can then check this flag in the scheduler ISR: if the
flag is set, then there is still a task running, and we have an “overload”
situation. We can report this using a global error variable and the
“Heartbeat” task that was introduced in Section 2.13.
Please note that this mechanism is intended primarily as a guide to the
system loading for use during development, but it can be included in
production systems (and, perhaps, checked during scheduled maintenance
sessions), if the task overrun situation is a “nuisance” indicator, rather than a
safety indicator. This may be the case in systems that have “soft” timing
constraints.
31
Please also note that this indicator clearly won’t have a chance to work if the
WDT setting in your system are set to match the tick interval (as in Figure
19).
You will therefore need to increase the WDT timeout settings if you intend
to use this mechanism (during development or in a production system).
TTRD02b includes a complete implementation of the overload detection
mechanism.
Code Fragment 9 shows the timer ISR function from TTRD02b: in this ISR
the “Task_running_G” flag is checked.
void SysTick_Handler(void)
{
// Increment tick count (only)
Tick_count_G++;
// As this is a TTC scheduler, we don't usually expect
// to have a task running when the timer ISR is called
if (Task_running_G == 1)
{
// Simple error reporting via heartbeat / error LED.
// (This value is *not* reset.)
Error_code_G = ERROR_SCH_SYSTEM_OVERLOAD;
}
}
Code Fragment 9: Detecting system overloads: Checking the “Task running” flag.
Code Fragment 10 shows how this flag can be set (when the task is released,
in the “Dispatcher” function).
// Check if there is a task at this location
if (SCH_tasks_G[Index].pTask)
{
if (--SCH_tasks_G[Index].Delay == 0)
{
// The task is due to run
// Set "Task_running" flag
__disable_irq();
Task_running_G = 1;
__enable_irq();
Code Fragment 10: Detecting system overloads: Setting the “Task running” flag.
2.15. Example: Injected (transitory) task overrun (TTRD02b)
We’ve introduced two simple mechanisms for detecting task overruns in this
chapter. In this section, we introduce a simple example that can be used to
test these mechanisms. The complete example is illustrated in TTRD02b.
32
Please refer to Listing 12. This shows part of a HEARTBEAT_Update task
has been adapted to generate a transient overrun event (of duration less than
4 ms), after 15 seconds.
Note that – with the standard WDT settings – this task will only overrun
once (because the overrun will be detected by means of the WDT, and the
system will be reset: after the reset, the system will enter a fail silent mode).
However, if we extend the WDT setting (to around 10 ms), we can use the
mechanisms introduced in Section 2.14 to detect (and report) the system
overload situation. These WDT setting are illustrated in TTRD02b.
2.16. Task overruns may not always be “A Bad Thing”
A “soft” TTC design may be able to tolerate occasional task overruns.
In some cases, we can go beyond this. Consider, for example, the design
illustrated in Figure 22.
A
B
E
F
G
H
...
Time
Figure 22: A system design in which there is a “theoretical” system overload.
In this design, Task A and Task B are both released in the first tick, but their
combined execution time significantly exceeds the tick interval.
In this design (and in many practical cases), this “theoretical” system
overload has no impact on the task schedule, because no tasks are due to be
released in the second tick. Such a design may well be considered
acceptable.
As we will see in Chapter 11, this type of task schedule forms the basis of
TTH and TTP scheduler architectures, both of which are in widespread use.
Note that – if you opt to run with longer task combinations in one or more
ticks – you may need to adjust the WDT settings for the system (or at least
for this system mode), in order to avoid WDT-related resets.
2.17. Porting the scheduler (TTRD02c)
All of the schedulers presented so far in this chapter have employed the
SysTick timer to generate the system Tick. Such a timer is common across
many microcontrollers based on an ARM core and the code can therefore be
easily ported.
When you use this approach, you should bear in mind that this solution was
intended (by ARM) to be used to generate a 10 ms tick, as is commonly
required in basic operating systems.
33
As SysTick is based on a 24-bit timer, the maximum interval is (224-1) ticks:
at 100 MHz (the SystemCoreClock frequency used in the majority of the
examples in this book), this provides a maximum tick interval of
(16,777,215 / 100,000,000) seconds, or approximately 160 ms.
As an example of an alternative way of generating system ticks, TTRD02c
illustrates the use of Timer 0 in the LPC1769 device to drive a TTC
scheduler. In this design, the required tick interval is provided in
microseconds, and the maximum tick interval is ((232 -1) / 100,000,000)
seconds, or approximately 42 seconds.
Key parts of the scheduler initialisation function for TTRD02c are shown in
Listing 13.
2.18. Conclusions
In this chapter, we’ve introduced some simple but flexible task schedulers
for use with sets of periodic co-operative tasks.
In Chapter 3, we present an initial case study that employs one of these
schedulers.
34
2.19. Code listings (TTRD02a)
/*--------------------------------------------------------------------*main.h (Release 2014-11a)
-------------------------------------------------------------------This is the Project Header file.
-*--------------------------------------------------------------------*/
#ifndef _MAIN_H
#define _MAIN_H 1
// Links
#include
#include
#include
#include
to target libraries
<lpc17xx.h>
<cr_section_macros.h>
<NXP/crp.h>
<lpc17xx_gpio.h>
// Required system operating frequency (in Hz)
// Will be checked in the scheduler initialisation file
#define Required_SystemCoreClock (100000000)
// System header
#include "../system/system_1769_001-0_c02a.h"
// Scheduler header
#include "../scheduler/ttc_sch_1769_001-0_c02a.h"
// Port header
#include "../port/port_1769_001-0_c02a.h"
//-----------------------------------------------------------------// System error codes
//-----------------------------------------------------------------// Scheduler error codes
#define ERROR_SCH_TOO_MANY_TASKS (1)
#define ERROR_SCH_ONE_SHOT_TASK (2)
// Other error codes may be added here, if required
//-----------------------------------------------------------------// SHOULD NOT GENERALLY NEED TO EDIT THE SECTION BELOW
//-----------------------------------------------------------------#define RETURN_NORMAL 0
#define RETURN_ERROR 1
#endif
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 1: TTRD02a (main.h).
35
/*------------------------------------------------------------------*port_1769_001-0_c02a.h (Release 2014-11a)
-----------------------------------------------------------------This is the "Port Header" file: it documents usage of port pins
in the project.
-*------------------------------------------------------------------*/
#ifndef _PORT_H
#define _PORT_H 1
// Project header
#include "../main/main.h"
// Heartbeat LED
// Connected to "LED2" on LPC1769 board
// Port 0, Pin 22
#define HEARTBEAT_LED_PORT (0)
#define HEARTBEAT_LED_PIN (0b10000000000000000000000)
// Add jumper wire on baseboard to control WDT
// WDT is enabled *only* if jumper is in place.
// (Jumper is read at init phase only)
// Port 2, Pin 3
#define WDT_JUMPER_PORT (2)
#define WDT_JUMPER_PIN (0b1000)
#endif
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 2: TTRD02a (port_1769_001-0_c02a.h).
36
/*--------------------------------------------------------------------*main.c (Release 2014-11a)
-------------------------------------------------------------------main file for TT project.
See _readme.txt for project information.
-*--------------------------------------------------------------------*/
// Project header
#include "main.h"
/*------------------------------------------------------------------*/
int main(void)
{
// Check mode, add tasks to schedule
SYSTEM_Init();
// Start the scheduler
SCH_Start();
while(1)
{
SCH_Dispatch_Tasks();
}
return 1;
}
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 3: TTRD02a (main.c).
37
/*------------------------------------------------------------------*ttc_sch_1769_001-0_c02a.h (Release 2014-11a)
-----------------------------------------------------------------See ttc_sch_1769_001-0_c02a.c for details.
-*------------------------------------------------------------------*/
#ifndef _SCHEDULER_H
#define _SCHEDULER_H 1
#include "../main/main.h"
// ------ Public data type declarations ---------------------------// User-define type to store required data for each task
typedef struct
{
// Pointer to the task (must be a 'void (void)' function)
void (*pTask) (void);
// Delay (ticks) until the task will (next) be run
// - see SCH_Add_Task() for further details
uint32_t Delay;
// Interval (ticks) between subsequent runs.
// - see SCH_Add_Task() for further details
uint32_t Period;
// Worst-case execution time (microseconds)
uint32_t WCET;
// Best-case execution time (microseconds)
uint32_t BCET;
// Incremented (by scheduler) when task is due to execute
uint32_t RunMe;
} sTask;
Listing 4: TTRD02a (ttc_sch_1769_001-0_c02a.h) [Part 1 of 2]
38
// ------ Public function prototypes ------------------------------void
void
void
SCH_Init(const uint32_t TICKms);
SCH_Start(void);
SCH_Dispatch_Tasks(void);
uint32_t SCH_Add_Task(void (* pTask)(),
const uint32_t DELAY,
const uint32_t PERIOD,
const uint32_t WCET,
const uint32_t BCET
);
// Tick interval (ms)
//
//
//
//
Offset (in Ticks)
Ticks
Microseconds
Microseconds
// ------ Public constants ----------------------------------------// The maximum number of tasks required at any one time
// during the execution of the program
//
// MUST BE CHECKED FOR EACH PROJECT (*not* dynamic)
#define SCH_MAX_TASKS (20)
//-----------------------------------------------------------------// Error codes for scheduler
//-----------------------------------------------------------------#define
#define
#define
#define
ERROR_SCH_TOO_MANY_TASKS (1)
ERROR_SCH_ONE_SHOT_TASK (2)
ERROR_SCH_CANNOT_DELETE_TASK (3)
ERROR_SCH_SYSTEM_OVERLOAD (4)
#endif
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 4 TTRD02a (ttc_sch_1769_001-0_c02a.h) [Part 2 of 2]
39
/*--------------------------------------------------------------------*ttc_sch_1769_001-0_c02a.c (Release 2014-11a)
-------------------------------------------------------------------Time-Triggered Co-operative (TTC) task scheduler for LPC1769.
See "The Engineering of Reliable Embedded Systems" (Chapter 2)
for further information about this scheduler.
-*--------------------------------------------------------------------*/
// Project header
#include "../main/main.h"
// ------ Public variable definitions -----------------------------// Used to report errors, if required, using Heartbeat / Error LED
// See Heartbeat task (if used) for basic error-reporting mechanism
// See Scheduler Header for details of error codes
// See Port Header for details of the error pin
uint32_t Error_code_G;
// ------ Private variable definitions ----------------------------// The array of tasks
// Check array size in scheduler header file
sTask SCH_tasks_G[SCH_MAX_TASKS];
// The current tick count
static volatile uint32_t Tick_count_G = 0;
// ------ Private function prototypes -----------------------------static void SCH_Go_To_Sleep(void);
void SysTick_Handler(void);
Listing 5: TTRD02a (ttc_sch_1769_001-0_c02a.c) [Part 1 of 7]
40
/*------------------------------------------------------------------*SCH_Init()
Scheduler initialisation function. Prepares scheduler
data structures and sets up timer interrupts every TICKms
milliseconds.
You must call this function before using the scheduler.
[Required_SystemCoreClock frequency can be found in main.h.]
-*------------------------------------------------------------------*/
void SCH_Init(const uint32_t TICKms)
{
uint32_t i;
// Reset the global error variable
Error_code_G = 0;
for (i = 0; i < SCH_MAX_TASKS; i++)
{
SCH_tasks_G[i].pTask = 0;
}
// Using CMSIS
//
//
//
//
//
if
Must check board oscillator frequency, etc
- see "system_lpc17xx.c" (in linked CMSIS project)
*If* these values have been set correctly for your hardware
SystemCoreClock gives the system operating frequency (in Hz)
(SystemCoreClock != Required_SystemCoreClock)
{
// Fatal error
SYSTEM_Perform_Safe_Shutdown();
}
// Now to set up SysTick timer for "ticks" at interval TICKms
if (SysTick_Config(TICKms * SystemCoreClock / 1000))
{
// Fatal error
SYSTEM_Perform_Safe_Shutdown();
}
// Timer is started by SysTick_Config():
// we need to disable SysTick timer and SysTick interrupt until
// all tasks have been added to the schedule.
SysTick->CTRL &= 0xFFFFFFFC;
}
Listing 5: TTRD02a (ttc_sch_1769_001-0_c02a.c) [Part 2 of 7]
41
/*------------------------------------------------------------------*SCH_Start()
Starts the scheduler, by enabling SysTick interrupt.
NOTE: Usually called after all regular tasks are added,
to keep the tasks synchronised.
NOTE: ONLY THE SCHEDULER INTERRUPT SHOULD BE ENABLED!!!
-*------------------------------------------------------------------*/
void SCH_Start(void)
{
// Enable SysTick timer
SysTick->CTRL |= 0x01;
// Enable SysTick interrupt
SysTick->CTRL |= 0x02;
}
/*------------------------------------------------------------------*SysTick_Handler()
[Function name determined by CMIS standard.]
This is the scheduler ISR. It is called at a rate
determined by the timer settings in the SCH_Init() function.
-*------------------------------------------------------------------*/
void SysTick_Handler(void)
{
// Increment tick count (only)
Tick_count_G++;
}
Listing 5: TTRD02a (ttc_sch_1769_001-0_c02a.c) [Part 3 of 7]
42
/*------------------------------------------------------------------*SCH_Dispatch_Tasks()
This is the 'dispatcher' function. When a task (function)
is due to run, SCH_Dispatch_Tasks() will run it.
This function must be called (repeatedly) from the main loop.
-*------------------------------------------------------------------*/
void SCH_Dispatch_Tasks(void)
{
uint32_t Index;
uint32_t Update_required = 0;
__disable_irq(); // Protect shared resource (Tick_count_G)
if (Tick_count_G > 0)
{
Tick_count_G--;
Update_required = 1;
}
__enable_irq();
while (Update_required)
{
// Go through the task array
for (Index = 0; Index < SCH_MAX_TASKS; Index++)
{
// Check if there is a task at this location
if (SCH_tasks_G[Index].pTask)
{
if (--SCH_tasks_G[Index].Delay == 0)
{
(*SCH_tasks_G[Index].pTask)(); // Run the task
// All tasks are periodic in this design
// - schedule task to run again
SCH_tasks_G[Index].Delay = SCH_tasks_G[Index].Period;
}
}
}
__disable_irq();
if (Tick_count_G > 0)
{
Tick_count_G--;
Update_required = 1;
}
else
{
Update_required = 0;
}
__enable_irq();
}
SCH_Go_To_Sleep();
}
Listing 5: TTRD02a (ttc_sch_1769_001-0_c02a.c) [Part 4 of 7]
43
/*------------------------------------------------------------------*SCH_Add_Task()
Causes a task (function) to be executed at regular intervals.
pTask
- The name of the task (function) to be scheduled.
NOTE: All scheduled functions must be 'void, void' that is, they must take no parameters, and have
a void return type (in this design).
DELAY
- The interval (Ticks) before the task is first executed.
PERIOD - Task period (in Ticks).
Must be > 0.
WCET
- Worst-Case Execution Time (microseconds)
[Used only for documentation in this design.]
BCET
- Best-Case Execution Time (microseconds)
[Used only for documentation in this design.]
RETURN VALUE:
Returns the position in the task array at which the task has been
added. If the return value is SCH_MAX_TASKS then the task could
not be added to the array (there was insufficient space, or the
requested task period was 0).
If the return value is < SCH_MAX_TASKS, then the task was added
successfully.
Note: this return value may be used (in later designs) to
support the use of backup tasks.
-*------------------------------------------------------------------*/
uint32_t SCH_Add_Task(void (* pTask)(),
const uint32_t DELAY,
const uint32_t PERIOD,
const uint32_t WCET,
const uint32_t BCET
)
{
uint32_t Return_value = 0;
uint32_t Index = 0;
// First find a gap in the array (if there is one)
while ((SCH_tasks_G[Index].pTask != 0) && (Index < SCH_MAX_TASKS))
{
Index++;
}
Listing 5: TTRD02a (ttc_sch_1769_001-0_c02a.c) [Part 5 of 7]
44
// Have we reached the end of the list?
if (Index == SCH_MAX_TASKS)
{
// Task list is full
//
// Set the global error variable
Error_code_G = ERROR_SCH_TOO_MANY_TASKS;
// Also return an error code
Return_value = SCH_MAX_TASKS;
}
// Check for "one shot" tasks
// - not permitted in this design
if (PERIOD == 0)
{
// Set the global error variable
Error_code_G = ERROR_SCH_ONE_SHOT_TASK;
// Also return an error code
Return_value = SCH_MAX_TASKS;
}
if (Return_value != SCH_MAX_TASKS)
{
// If we're here, there is a space in the task array
// and the task to be added is periodic
SCH_tasks_G[Index].pTask = pTask;
SCH_tasks_G[Index].Delay
SCH_tasks_G[Index].Period
SCH_tasks_G[Index].WCET
SCH_tasks_G[Index].BCET
=
=
=
=
DELAY + 1;
PERIOD;
WCET;
BCET;
Return_value = Index;
}
return Return_value;
}
Listing 5: TTRD02a (ttc_sch_1769_001-0_c02a.c) [Part 6 of 7]
45
/*------------------------------------------------------------------*SCH_Go_To_Sleep()
This scheduler enters 'sleep mode' between clock ticks
to [i] reduce Tick jitter; and [ii] save power.
The next clock tick will return the processor
to the normal operating state.
Note: a slight performance improvement may be achieved
if this code is pasted into the 'Dispatch' function
(this may be at the cost of code readability & portability)
*** May be able to make further improvements to the jitter
*** behaviour depending on the target hardware architecture
*** Various power-saving options can be added
*** (e.g. shut down unused peripherals)
-*------------------------------------------------------------------*/
void SCH_Go_To_Sleep()
{
// Enter sleep mode = "Wait For Interrupt"
__WFI();
}
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 5: TTRD02a (ttc_sch_1769_001-0_c02a.c) [Part 7 of 7]
46
/*------------------------------------------------------------------*system_1769_001-0_c02a.h (Release 2014-11a)
-----------------------------------------------------------------See system_1769_001-0_c02a.c for details.
-*------------------------------------------------------------------*/
#ifndef _SYSTEM_H
#define _SYSTEM_H 1
// Two possible system modes
typedef enum {FAIL_SILENT, NORMAL} eSystem_mode;
// ------ Public function prototypes ------------------------------void SYSTEM_Init(void);
void SYSTEM_Perform_Safe_Shutdown(void);
#endif
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 6: TTRD02a (system_1769_001-0_c02a.h)
47
/*--------------------------------------------------------------------*system_1769_001-0_c02a.c (Release 2014-11a)
Controls system configuration after processor reset.
[Two modes supported - "Normal" and "Fail Silent".]
-*--------------------------------------------------------------------*/
// Project header
#include "../main/main.h"
// Task headers
#include "../tasks/heartbeat_1769_001-0_c02a.h"
#include "../tasks/watchdog_1769_001-0_c02a.h"
// ------ Public variable definitions -----------------------------// In many designs, System_mode_G will be used in other modules.
// - we therefore make this variable public.
eSystem_mode System_mode_G;
// ------ Private function declarations ---------------------------void SYSTEM_Identify_Required_Mode(void);
void SYSTEM_Configure_Required_Mode(void);
/*------------------------------------------------------------------*SYSTEM_Init()
Wrapper for system startup functions.
-*------------------------------------------------------------------*/
void SYSTEM_Init(void)
{
SYSTEM_Identify_Required_Mode();
SYSTEM_Configure_Required_Mode();
}
Listing 7: TTRD02a (system_1769_001-0_c02a.c) [Part 1 of 4]
48
/*------------------------------------------------------------------*SYSTEM_Identify_Required_Mode()
Try to work out the cause of the system reset.
Set the system mode accordingly.
-*------------------------------------------------------------------*/
void SYSTEM_Identify_Required_Mode(void)
{
uint32_t wd_flag;
// If "1", reset was caused by WDT
wd_flag = (LPC_SC->RSID >> 2) & 1;
if (wd_flag == 1)
{
// Cleared only by software or POR
// Clear flag (or other resets may be interpreted as WDT)
LPC_SC->RSID &= ~(0x04);
// Set system mode (Fail Silent)
System_mode_G = FAIL_SILENT;
}
else
{
// Here we treat all other forms of reset in the same way
// Set system mode (Normal)
System_mode_G = NORMAL;
}
}
Listing 7: TTRD02a (system_1769_001-0_c02a.c) [Part 2 of 4]
49
/*------------------------------------------------------------------*SYSTEM_Configure_Required_Mode()
Configure the system in the required mode.
-*------------------------------------------------------------------*/
void SYSTEM_Configure_Required_Mode(void)
{
switch (System_mode_G)
{
case default:
// Default to "FAIL_SILENT"
case FAIL_SILENT:
{
// Reset caused by WDT
// Trigger "fail silent" behaviour
SYSTEM_Perform_Safe_Shutdown();
break;
}
case NORMAL:
{
// Set up WDT (timeout in *microseconds*)
WATCHDOG_Init(1100);
// Set up scheduler for 1 ms ticks (tick interval in *ms*)
SCH_Init(1);
// Prepare for heartbeat task
HEARTBEAT_Init();
//
//
//
//
//
//
//
Add tasks to schedule.
Parameters are:
1. Task name
2. Initial delay / offset (in Ticks)
3. Task period (in Ticks): Must be > 0
4. Task WCET (in microseconds)
5. Task BCET (in microseconds)
// Add watchdog task first
SCH_Add_Task(WATCHDOG_Update, 0, 1, 10, 0);
// Add heartbeat task
SCH_Add_Task(HEARTBEAT_Update, 0, 1000, 20, 0);
// Feed the watchdog
WATCHDOG_Update();
break;
}
}
}
Listing 7: TTRD02a (system_1769_001-0_c02a.c) [Part 3 of 4]
50
/*------------------------------------------------------------------*SYSTEM_Perform_Safe_Shutdown()
Attempt to place the system into a safe state.
Note: Does not return and may (if watchdog is operational) result
in a processor reset, after which the function may be called again.
[The rational for this behaviour is that - after the reset the system MAY be in a better position to enter a safe state.
To avoid the possible reset, adapt the code and feed the WDT
in the loop.]
-*------------------------------------------------------------------*/
void SYSTEM_Perform_Safe_Shutdown(void)
{
// Used for simple error reporting
uint32_t Delay, j;
// Here we simply "fail silent" with rudimentary error reporting
// OTHER BEHAVIOUR IS LIKELY TO BE REQUIRED IN YOUR DESIGN
// *************************************
// NOTE: This function should NOT return
// *************************************
HEARTBEAT_Init();
while(1)
{
// Flicker Heartbeat LED to indicate error
for (Delay = 0; Delay < 200000; Delay++) j *= 3;
HEARTBEAT_Update();
}
}
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 7: TTRD02a (system_1769_001-0_c02a.c) [Part 4 of 4]
51
/*--------------------------------------------------------------------*heartbeat_1769_001-0_c02a.h (Release 2014-11a)
-------------------------------------------------------------------- See heartbeat_1769_001-0_c02a.c for details.
-*--------------------------------------------------------------------*/
#ifndef _HEARTBEAT_H
#define _HEARTBEAT_H 1
// ------ Public function prototypes ------------------------------void HEARTBEAT_Init(void);
void HEARTBEAT_Update(void);
#endif
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 8: TTRD02a (heartbeat_1769_001-0_c02a.h)
52
/*--------------------------------------------------------------------*heartbeat_1769_001-0_c02a.c (Release 2014-11a)
-------------------------------------------------------------------Simple 'Heartbeat' task for LPC1769.
If everything is OK, flashes at 0.5 Hz
If there is an error code active, this is displayed.
-*--------------------------------------------------------------------*/
// Project header
#include "../main/main.h"
// Task header
#include "heartbeat_1769_001-0_c02a.h"
// ------ Public variable declarations ----------------------------// See ttc_scheduler_basic.c for definition
extern uint32_t Error_code_G;
/*------------------------------------------------------------------*HEARTBEAT_Init()
Prepare for HEARTBEAT_Update() function - see below.
-*------------------------------------------------------------------*/
void HEARTBEAT_Init(void)
{
// Set up LED2 as an output pin
// Params: Port : Pin : 1 for o/p, 0 for i/p
GPIO_SetDir(HEARTBEAT_LED_PORT, HEARTBEAT_LED_PIN, 1);
}
Listing 9: TTRD02a (heartbeat_1769_001-0_c02a.c) [Part 1 of 3]
53
/*------------------------------------------------------------------*HEARTBEAT_Update()
Flashes at 0.5 Hz if error code is 0 (i.e. no error code).
Otherwise, displays error code.
Must schedule every second (soft deadline).
-*------------------------------------------------------------------*/
void HEARTBEAT_Update(void)
{
static uint32_t Heartbeat_state = 0;
static uint32_t Error_state = 0;
if (Error_code_G == 0)
{
// No errors recorded
// - just flash at 0.5 Hz
// Change the LED from OFF to ON (or vice versa)
if (Heartbeat_state == 1)
{
Heartbeat_state = 0;
GPIO_ClearValue(HEARTBEAT_LED_PORT, HEARTBEAT_LED_PIN);
}
else
{
Heartbeat_state = 1;
GPIO_SetValue(HEARTBEAT_LED_PORT, HEARTBEAT_LED_PIN);
}
}
Listing 9: TTRD02a (heartbeat_1769_001-0_c02a.h) [Part 2 of 3]
54
else
{
// If we are here, there is a (non-zero) error code ...
Error_state++;
if (Error_state < Error_code_G*2)
{
Heartbeat_state = 0;
GPIO_ClearValue(HEARTBEAT_LED_PORT, HEARTBEAT_LED_PIN);
}
else
{
if (Error_state < Error_code_G*4)
{
// Change the LED from OFF to ON (or vice versa)
if (Heartbeat_state == 1)
{
Heartbeat_state = 0;
GPIO_ClearValue(HEARTBEAT_LED_PORT, HEARTBEAT_LED_PIN);
}
else
{
Heartbeat_state = 1;
GPIO_SetValue(HEARTBEAT_LED_PORT, HEARTBEAT_LED_PIN);
}
}
else
{
Error_state = 0;
}
}
}
}
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 9: TTRD02a (heartbeat_1769_001-0_c02a.h) [Part 3 of 3]
55
/*------------------------------------------------------------------*watchdog_1769_001-0_c02a.h (Release 2014-11a)
------------------------------------------------------------------ See watchdog_1769_001-0_c02a.c for details.
-*------------------------------------------------------------------*/
#ifndef _WDT_H
#define _WDT_H 1
#include <lpc17xx_wdt.h>
// ------ Public constants ----------------------------------------// From NXP
/** Define divider index for microsecond ( us ) */
#define WDT_US_INDEX ((uint32_t)(1000000))
/** WDT Time out minimum value */
#define WDT_TIMEOUT_MIN ((uint32_t)(0xFF))
/** WDT Time out maximum value */
#define WDT_TIMEOUT_MAX ((uint32_t)(0xFFFFFFFF))
// Jumper connections
// WDT will only be enabled if jumper is inserted
// (connecting the jumper pin to ground)
// - see Port.H for jumper pin details.
#define WDT_JUMPER_INSERTED (0)
#define WDT_PCLK (4000000)
// ------ Public function prototypes ------------------------------void WATCHDOG_Init(const uint32_t);
void WATCHDOG_Update(void);
#endif
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 10: TTRD02a (watchdog_1769_001-0_c02a.h)
56
/*--------------------------------------------------------------------*watchdog_1769_001-0_c02a.c (Release 2014-11a)
-------------------------------------------------------------------'Watchdog' library for LPC1769.
** Jumper controlled (see below) **
-*--------------------------------------------------------------------*/
// Project header
#include "../main/main.h"
// Task header
#include "watchdog_1769_001-0_c02a.h"
// ------ Public variable declarations ----------------------------// See scheduler module for definition
extern uint32_t Error_code_G;
/*------------------------------------------------------------------*WATCHDOG_Init()
Set up watchdog timer on LPC1769.
*****************************************************************
* Handle with care - if WDT is running, debug links may be lost *
* In this design, WDT is enable only when jumper is inserted.
*
*****************************************************************
The watchdog timer is driven by the Internal RC Oscillator:
the minimum available timeout is 256 Вµs.
-*------------------------------------------------------------------*/
Listing 11: TTRD02a (watchdog_1769_001-0_c02a.c) [Part 1 of 3]
57
void WATCHDOG_Init(const uint32_t WDT_TIMEOUTus)
{
uint32_t wdt_ticks = 0;
uint32_t Jumper_input;
// *If* WDT jumper is in place, we start the WDT
// Read WDT jumper setting
// - set up jumper pin for input
// - params: Port : Pin : 1 for o/p, 0 for i/p
GPIO_SetDir(WDT_JUMPER_PORT, WDT_JUMPER_PIN, 0);
// Note: we only read the jumper during system init phase
Jumper_input = (GPIO_ReadValue(WDT_JUMPER_PORT) & WDT_JUMPER_PIN);
if (Jumper_input != WDT_JUMPER_INSERTED)
{
// Jumper not inserted - don't enable WDT
return;
}
// If we are here, we are setting up the WDT
// Drive WDT from internal RC timer (IRC)
LPC_WDT->WDCLKSEL = 0x00;
// Calculate required tick count for WDT timeout
wdt_ticks = (((WDT_PCLK) / WDT_US_INDEX) * (WDT_TIMEOUTus / 4));
// Check if tick count is within allowed range
if ((wdt_ticks >= WDT_TIMEOUT_MIN) && (wdt_ticks <= WDT_TIMEOUT_MAX))
{
LPC_WDT->WDTC = wdt_ticks;
}
else
{
// We simply "stop" if WDT values are wrong
// - other solutions may make sense for your application
// - for example, use closest available timeout.
while(1);
}
// Reset if WDT overflows
LPC_WDT->WDMOD |= 0x02;
// Start WDT
LPC_WDT->WDMOD |= 0x01;
// Feed watchdog
WATCHDOG_Update();
}
Listing 11: TTRD02a (watchdog_1769_001-0_c02a.c) [Part 2 of 3]
58
/*------------------------------------------------------------------*WATCHDOG_Update()
Feed the watchdog timer.
See Watchdog_Init() for further information.
-*------------------------------------------------------------------*/
void WATCHDOG_Update(void)
{
// Feed the watchdog
__disable_irq(); // Avoid possible interruption
LPC_WDT->WDFEED = 0xAA;
LPC_WDT->WDFEED = 0x55;
__enable_irq();
}
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 11: TTRD02a (watchdog_1769_001-0_c02a.c) [Part 3 of 3]
59
2.20. Code listings (TTRD02b)
/*------------------------------------------------------------------*HEARTBEAT_Update()
Flashes at 0.5 Hz if error code is 0 (i.e. no error code).
Otherwise, displays error code.
* Incorporates fault injection (task overrun) after 15 secs *
Must schedule every second (soft deadline).
-*------------------------------------------------------------------*/
void HEARTBEAT_Update(void)
{
static uint32_t Heartbeat_state = 0;
static uint32_t Error_state = 0;
uint32_t Delay, j;
static uint32_t Task_overrun_counter = 0;
// Force task overrun after 15 seconds (test / demo purposes)
if (Task_overrun_counter++ == 15)
{
Task_overrun_counter = 0;
// Trigger temporary task overrun (for demo purposes)
// This gives delay of ~3.6 ms
for (Delay = 0; Delay < 20000; Delay++)
{
j *= 3;
}
}
if (Error_code_G == 0)
{
// No errors recorded
// - just flash at 0.5 Hz
// Remaining code omitted here
...
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 12: TTRD02b (extract from file heartbeat_1769_001-0_c02b.c)
60
2.21. Code listings (TTRD02c)
void SCH_Init(const uint32_t TICKmicroseconds)
{
// Used to configure Timer 0
TIM_TIMERCFG_Type TMR0_Cfg;
TIM_MATCHCFG_Type TMR0_Match;
// Code removed
// Initialise Timer 0, prescale counter = 1us
TMR0_Cfg.PrescaleOption = TIM_PRESCALE_USVAL;
TMR0_Cfg.PrescaleValue = 1;
// Use channel 0, MR0
TMR0_Match.MatchChannel = 0;
// Enable interrupt when MR0 matches the value in TC register
TMR0_Match.IntOnMatch = ENABLE;
// Enable reset on MR0: TIMER will reset if MR0 matches it
TMR0_Match.ResetOnMatch = TRUE;
// Don't stop on MR0 if MR0 matches it
TMR0_Match.StopOnMatch = FALSE;
// Do nothing for external output pin if match
TMR0_Match.ExtMatchOutputType = TIM_EXTMATCH_NOTHING;
// Tick value
// Set Match value, count value in microseconds in this version.
TMR0_Match.MatchValue = TICKmicroseconds;
// Set configuration for Tim_config and Tim_MatchConfig
TIM_Init(LPC_TIM0, TIM_TIMER_MODE, &TMR0_Cfg);
TIM_ConfigMatch(LPC_TIM0, &TMR0_Match);
// Highest priority = Timer 0
NVIC_SetPriority(TIMER0_IRQn, 0);
}
Listing 13: TTRD02c (extract from file ttc_sch_od_1769_001-0_c02c.c)
61
CHAPTER 3: Initial case study
In this chapter we consider present an introductory case study.
Related TTRDs
A list of the TTRDs discussed in this chapter is included below.
C programming language (LPC1769 target):
 TTRD03a – Simple TTC control system for washing machine
3.1. Introduction
As noted in the Preface, the goal of book is to present a model-based process
for the development of embedded applications that can be used to provide
evidence that the system concerned will be able to detect System Faults (as
defined on Page xv) and then handle them in an appropriate manner, thereby
avoiding Uncontrolled System Failures.
Before we introduce the model-based development process, we will present
a simple case study in this chapter.
3.2. The focus of this case study
The examples in this book target the LPC1769 microcontroller. As noted in
the Preface, this processor is intended for use in applications such as:
industrial networking; motor control; white goods; eMetering; alarm
systems; and lighting control.
This case study will focus on the use of the LPC1769 in a controller for a
domestic washing machine. This is an appropriate application for this
microcontroller. However, the TT software architecture that is employed is
generic in nature and can also be employed in many other systems.
3.3. The purpose of this case study
We will present a simple TT framework for a control system in this chapter.
In subsequent chapters of this book, we will present a range of techniques
that will allow us to make this framework much more robust in the presence
of various potential threats (such as EMI).
While the framework created here could form the foundation for a reliable
embedded system, the initial design presented in this chapter is far from
complete and is best viewed as an early system prototype.
We will revisit this study in Chapter 15 and apply the techniques described
in the remainder of this book in order to improve the system reliability.
63
Washer
controller
Start
switch
Water
valve
Selector
dial
Water
heater
Water
pump
Door-position
sensor
Washer
controller
Drum
motor
Water-level
sensor
Detergent
hatch
Watertemperature
sensor
[a]
[b]
LED
indicators
[h]
[c]
[i]
[j]
[k]
[d]
[e]
[l]
[f]
[g]
Figure 23: An overview of the sensor and actuator components in the washer control system.
Internal view of the system (bottom left): [a] water value; [b] detergent hatch; [c] water pump;
[d] water-level sensor; [e] water-temperature sensor; [f] drum motor; [g] water heater.
External view of the system (bottom right): [h] selector dial; [i] LED indicators; [j] start switch;
[k] door lock; [l] door-position sensor.
64
3.4. A summary of the required system operation
An overview of the sensor and actuator components in the washer control
system is given in Figure 23.
This is a brief summary of the required system operation:
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
The user selects a wash program (e.g. �Cotton’) on the selector dial.
The user presses the �Start’ switch.
The door lock is engaged.
The water valve is opened to allow water into the wash drum.
If the wash program involves detergent, the detergent hatch is opened.
When the detergent has been released, the detergent hatch is closed.
When the �full water level’ is sensed, the water valve is closed.
If the program involves warm water, the water heater is switched on.
When the water reaches the correct temperature, the water heater is
switched off.
The washer motor is turned on to rotate the drum. The motor then goes
through a series of movements, both forward and reverse (at various
speeds) to wash the clothes. (The precise set of movements carried out
depends on the wash program that the user has selected).
At the end of the wash cycle, the motor is stopped.
The pump is switched on to drain the drum.
When the drum is empty, the pump is switched off.
The door lock is released.
During the operation various LEDs are used to indicate where the
system is in the wash cycle.
The description is simplified for the purposes of this example, but it will be
adequate for our purposes here.
Please note that we do not use the door-position sensor in this version of the
case study (we will use this in Chapter 15).
3.5. The system architecture
Our design consists of a “system” task, plus one module for each
component11 on the context diagram shown in Figure 23.
The system task stores and updates the current state. It operates by reading
information from the various sensor tasks and controlling the system
operation by means of the various actuator tasks.
The underlying architecture is widely applicable. For example, the sequence
of events used to raise the landing gear in a passenger aircraft can be
controlled in a similar manner. In this case, basic tests (such as �WoW’ –
11
With the exception of the door-position sensor that is not used in this version of the study.
65
�Weight on Wheels’) will be used to determine whether the aircraft is on the
ground or in the air: these tests will be completed before the operation
begins. Feedback from various door and landing-gear sensors will then be
used to ensure that each phase of the manoeuvre completes correctly.
3.6. The system states
The possible system states are described in Table 2.
The various states are implemented by means of a switch statement in the
main “washer controller” task (see Section 3.8).
3.7. Implementation platform for the prototype
The prototype system described in this chapter is designed to run on the EA
Baseboard: this is described in Appendix 2.
The key user-interface elements in the prototype are illustrated
schematically in Figure 24.
[a]
[b]
[c]
[d]
Figure 24: User-interface elements in the washer prototype: [a] RGB LED cluster indicating the
state of the motor and door lock [b] 7-segment display indicating the overall system state;
[c] the “heartbeat” LED; [d] connections needed for the WDT (as detailed in Chapter 2).
In this design, the overall state of the system is illustrated by the 7-segment
LED. For example – as noted in the left column in Table 2 - “2” will be
displayed in the FILL_DRUM state, and “E” will be displayed in the
FAULT state.
The state of the motor is indicated by the blue LED in the RGB LED cluster.
The green LED in this cluster is on (and the red LED is off) when the door
is unlocked; the red LED is on (and the green LED is off) when the door is
locked.
There is a standard “heartbeat” LED, and the WDT support requires a
jumper to be inserted (as detailed in Chapter 2, Section 2.11).
66
Table 2: Possible system states for the washing-machine controller.
State
Description
INIT
The system enters INIT state when it is powered up. It remains in this
state until the Start switch is pressed. Once the Start switch is pressed, the
system reads the selector dial: the system then moves into the START
state.
[0]
START
[1]
FILL_DRUM
[2]
HEAT_WATER
[3]
PRE_WASH
[4]
MAIN_WASH
[5]
DRAIN_DRUM
[6]
SPIN
[7]
FINISHED
[-]
FAULT
[E]
In the START state, the door is locked, and the water valve is opened to
begin filling the drum with cold water. If detergent is required (depending
on the reading from the selector dial) then this will also be released into
the drum. The system then enters FILL_DRUM state.
In FILL_DRUM state, the system checks the water-level sensor periodically,
to see if the drum is full. If it takes longer than the MAX_FILL_DURATION
to fill the drum, then something is wrong (for example, there is no water
supply connected): the system then enters FAULT state. If the drum has
filled in time, then the water heater may be switched on (if hot water is
required in the selected wash routine). If the water is required, the system
will move into HEAT_WATER state. If hot water is not required, the system
will move into PRE_WASH state.
In HEAT_WATER state, the system checks the water-temperature sensor
periodically, to see if the water has reached the required temperature. If it
takes longer than the MAX_WATER_HEAT_DURATION to heat the water,
then something is wrong (for example, the water heater has failed): the
system then enters FAULT state. If the water reaches the required
temperature within the time limit then the system will move into
PRE_WASH state.
In the PRE_WASH state, the drum motor is activated in order to rotate the
drum. In our simplified design, this will simply involve running the motor
(constantly) at “medium speed”. In a more complete design, we would
expect to perform a more complicated sequence of moves in this state (for
example, changing the speed and direction of rotation): such features
could be added to the present design without great difficulty. In this
example, the system remains in the pre-wash state for a period equal to
the PRE_WASH_DURATION. It then moves into the MAIN_WASH state.
Like the PRE_WASH state, the MAIN_WASH state involves turning the
drum (in this case at a higher speed). We remain in this state for a period
equal to the MAIN_WASH_DURATION. The system then moves into the
DRAIN_DRUM state.
In the DRAIN_DRUM state, the water pump is activated: this begins the
process of removing water from the drum. If draining the drum takes
longer than DRAIN_DRUM_DURATION, we move into the FAULT state,
otherwise – after the drum has been drained – we move into the SPIN
state.
In the SPIN state, the drum is rotated at full speed. We remain in this state
for a period equal to the SPIN_DURATION. The system then moves into
the FINISHED state.
In the FINISHED state, the various system components are switched off.
To avoid any risk of injury when the door is opened (just in case the drum
is still rotating), there is a delay (of 10 seconds) before the door lock is
released. The system then returns to the INIT state, ready for the next
wash.
In the event that a fault is detected, the system enters the FAULT state. All
components are switched off, and there is a delay before the door is
unlocked. The system remains (indefinitely) in the fault state.
67
3.8. The “system” task
The system task is shown in Listing 15.
The possible system states are implemented by means of an enumerated type
(Code Fragment 11).
// Possible system states
typedef enum {INIT, START, FILL_DRUM, HEAT_WATER, PRE_WASH,
MAIN_WASH, DRAIN_DRUM, SPIN, FINISHED, FAULT}
eSystem_state;
Code Fragment 11: The possible washer states.
Just in case a value outside the allowed range is assigned to the
System_state_G variable, we have a default state (which is FAIL_SILENT):
see Code Fragment 12.
switch (System_state_G)
{
case INIT: // 0
{
}
...
default:
// Default state
case FAULT: // E
{
}
}
Code Fragment 12: The washer default state.
Please note that use of default states is a requirement in MISRA C.
3.9. The “selector dial” task
In the final system, the selector-dial task will read an input from a switch
something like that shown in Figure 25. In our demo system, we use a
simple placeholder task.
Figure 25: A typical selector dial from a domestic washing machine.
68
3.10. The “start switch” task
In our demonstration system, we use “SW3” on the EA Baseboard as the
start switch.
3.11. The “door lock” task
In our demonstration system, the door lock is representation by the state of
two LEDs (as detailed in Section 3.7).
3.12. The “water valve” task
A simple placeholder task is used for the water-valve interface in this demo.
3.13. The “detergent hatch” task
A simple placeholder task is used for the detergent-hatch interface in this
demo.
3.14. The “water level” task
A simple model is used to implement the water-level task in the demo
system (Code Fragment 13).
3.15. The “water heater” task
A simple placeholder task is used for the water-heater interface in this demo.
3.16. The “water temperature” task
Another simple model is used to implement the water-temperature task in
this demo system.
3.17. The “drum motor” task
In our demonstration system, the drum-motor state is representation by the
state of the blue LED on the EA Baseboard (as detailed in Section 3.7).
3.18. The “water pump” task
A simple placeholder task is used for the water-pump interface in this demo.
3.19. The “heartbeat” task
A standard heartbeat LED is used in the demo (please refer again to
Section 3.7 for connection details).
69
void WATER_LEVEL_Update(void)
{
// Placeholder code
static uint32_t Water_level = 0;
// Is the drum filling?
if ((Water_valve_required_state_G == OPEN) && !WATER_VALVE_FAULT)
{
// Drum is filling
Water_level++;
// Is the drum full of water (ready to wash)?
if (Water_level >= 50)
{
Water_level_reading_G = 1;
}
}
// Is the drum being drained?
if ((Water_pump_required_state_G == ON) && !WATER_PUMP_FAULT)
{
// Drum is being drained
if (Water_level > 0)
{
// Drum is draining
Water_level--;
}
if (Water_level == 0)
{
// Drum is empty
Water_level_reading_G = 0;
}
}
}
Code Fragment 13: The water-level task, based on a simple model.
3.20. Communication between tasks
Communication between tasks is carried out by means of global variables.
For example, the required state of the water pump is represented as shown in
Code Fragment 14.
// ------ Public variable definitions -------------------------------uint32_t Water_pump_required_state_G;
Code Fragment 14: The global variable used to control the state of the water pump.
70
3.21. Where do we go from here?
In this case study we have illustrated how a set of “placeholder” tasks and a
TT scheduler can be employed in order to develop a framework for a nontrivial control system.
We emphasise again that – while the framework created here could form the
starting point for a reliable embedded system - the result so far (as detailed
in full in TTRD03a) is far from complete and is best viewed as an early
system prototype.
We will build on this foundation in later chapters, following a three-stage
development process:
п‚·
п‚·
п‚·
The first stage will involve modelling the system timing characteristics:
we will begin to explore this process in Chapter 4.
The second stage will involve building the system using appropriate task
designs (Chapter 6), along with a TT scheduler that has support for
“Task Contracts” (Chapter 8 and Chapter 9) and – if required – support
for task pre-emption (Chapter 11 and Chapter 12).
The third stage will involve adding comprehensive support for run-time
monitoring: we begin to see how this can be achieved in Chapter 8.
The end result will meet our requirements for a reliable embedded system
and will be applicable in a wide range of sectors.
As far as this particular study is concerned, two of our goals in the second
version of this system (presented in Chapter 15) will be to ensure that:
[i] the door of our washing machine cannot be opened while the drum is
spinning, an event that would risk injury to users of the product; [ii] the door
can be opened – without causing a flood – in circumstances where it is
realised that a piece of coloured clothing has been inadvertently included in
a white wash, or that a toddler has placed the family cat in the machine with
the washing.
It should be emphasised that very similar challenges apply in other sectors.
For example, in Chapter 1 (Section 1.8), we discussed a system that is
intended to engage the steering-column lock in a passenger vehicle (to
secure the vehicle against theft) after the driver has parked and left the
vehicle. In this system we need to ensure (for security reasons) that this
lock is engaged when the vehicle is parked: however, we also need to ensure
(for safety reasons) that this lock can never be engaged when the vehicle is
moving.
71
3.22. Conclusions
In this chapter, we have illustrated how a TTC scheduler can be used with
an initial set of periodic “placeholder” tasks in order to create a prototype
for an embedded control system.
In Chapter 4, we will begin to explore techniques for modelling sets of
periodic tasks in order to determine (for example) precisely how long it will
take our system to respond to external events.
72
3.23. Code listings (TTRD03a)
/*--------------------------------------------------------------------*system_1769_001-0_c03a.c (Release 2014-11a)
Controls system configuration after processor reset.
[Two modes supported - "Normal" and "Fail Silent".]
-*--------------------------------------------------------------------*/
// Project header
#include "../main/main.h"
// Task headers
#include "../tasks/detergent_hatch_1769_001-0_c03a.h"
#include "../tasks/door_lock_1769_001-0_c03a.h"
#include "../tasks/drum_motor_1769_001-0_c03a.h"
#include "../tasks/selector_dial_1769_001-0_c03a.h"
#include "../tasks/start_switch_1769_001-0_c03a.h"
#include "../tasks/washer_controller_1769_001-0_c03a.h"
#include "../tasks/water_heater_1769_001-0_c03a.h"
#include "../tasks/water_level_1769_001-0_c03a.h"
#include "../tasks/water_pump_1769_001-0_c03a.h"
#include "../tasks/water_temperature_1769_001-0_c03a.h"
#include "../tasks/water_valve_1769_001-0_c03a.h"
#include "../tasks/heartbeat_1769_001-0_c02a.h"
#include "../tasks/watchdog_1769_001-0_c02a.h"
// ------ Public variable definitions -----------------------------// In many designs, System_mode_G will be used in other modules.
// - we therefore make this variable public.
eSystem_mode System_mode_G;
// ------ Public variable declarations ----------------------------// Actuators
extern uint32_t
extern uint32_t
extern uint32_t
extern uint32_t
extern uint32_t
extern uint32_t
extern uint32_t
Drum_motor_required_state_G;
Drum_motor_required_speed_G;
Door_lock_required_state_G;
Detergent_hatch_required_state_G;
Water_heater_required_state_G;
Water_pump_required_state_G;
Water_valve_required_state_G;
// ------ Private function declarations ---------------------------void SYSTEM_Identify_Required_Mode(void);
void SYSTEM_Configure_Required_Mode(void);
Listing 14: TTRD03a (system_1769_001-0_c03a.c) [Part 1 of 5]
73
/*------------------------------------------------------------------*SYSTEM_Init()
Wrapper for system startup functions.
-*------------------------------------------------------------------*/
void SYSTEM_Init(void)
{
SYSTEM_Identify_Required_Mode();
SYSTEM_Configure_Required_Mode();
}
/*------------------------------------------------------------------*SYSTEM_Identify_Required_Mode()
Try to work out the cause of the system reset.
Set the system mode accordingly.
-*------------------------------------------------------------------*/
void SYSTEM_Identify_Required_Mode(void)
{
uint32_t wd_flag;
// If "1", reset was caused by WDT
wd_flag = (LPC_SC->RSID >> 2) & 1;
if (wd_flag == 1)
{
// Cleared only by software or POR
// Clear flag (or other resets may be interpreted as WDT)
LPC_SC->RSID &= ~(0x04);
// Set system mode (Fail Silent)
System_mode_G = FAIL_SILENT;
}
else
{
// Here we treat all other forms of reset in the same way
// Set system mode (Normal)
System_mode_G = NORMAL;
}
}
Listing 14: TTRD03a (system_1769_001-0_c03a.c) [Part 2 of 5]
74
/*------------------------------------------------------------------*SYSTEM_Configure_Required_Mode()
Configure the system in the required mode.
-*------------------------------------------------------------------*/
void SYSTEM_Configure_Required_Mode(void)
{
switch (System_mode_G)
{
case default:
// Default to "FAIL_SILENT"
case FAIL_SILENT:
{
// Reset caused by WDT
// Trigger "fail silent" behaviour
SYSTEM_Perform_Safe_Shutdown();
break;
}
Listing 14: TTRD03a (system_1769_001-0_c03a.c) [Part 3 of 5]
75
case NORMAL:
{
// Set up WDT (timeout in *microseconds*)
WATCHDOG_Init(1100);
// Set up scheduler for 1 ms ticks (tick interval in *ms*)
SCH_Init(1);
// Prepare for heartbeat task
HEARTBEAT_Init();
// Prepare to read START switch
SWITCH_SW3_Init();
// Prepare for main washer task
WASHER_CONTROLLER_Init();
// Prepare for other tasks
DETERGENT_HATCH_Init();
DOOR_LOCK_Init();
DRUM_MOTOR_Init();
SELECTOR_DIAL_Init();
WATER_HEATER_Init();
WATER_LEVEL_Init();
WATER_PUMP_Init();
WATER_TEMPERATURE_Init();
WATER_VALVE_Init();
//
//
//
//
//
//
//
Add tasks to schedule.
Parameters are:
1. Task name
2. Initial delay / offset (in Ticks)
3. Task period (in Ticks) - set to 0 for "one shot" task
4. Task WCET (in microseconds)
5. Task BCET (in microseconds)
// Add watchdog task first
SCH_Add_Task(WATCHDOG_Update, 0, 1, 10, 0);
// Add task to control door lock
SCH_Add_Task(DOOR_LOCK_Update, 0, 10, 100, 0);
// Add tasks to read selector dial, then START switch
SCH_Add_Task(SELECTOR_DIAL_Update, 0, 10, 100, 0);
SCH_Add_Task(SWITCH_SW3_Update, 0, 10, 100, 0);
// Remaining tasks called less frequently
SCH_Add_Task(DETERGENT_HATCH_Update, 1, 100, 100, 0);
SCH_Add_Task(DRUM_MOTOR_Update, 2, 100, 100, 0);
SCH_Add_Task(WATER_HEATER_Update, 3, 100, 100, 0);
SCH_Add_Task(WATER_LEVEL_Update, 4, 100, 100, 0);
SCH_Add_Task(WATER_PUMP_Update, 5, 100, 100, 0);
SCH_Add_Task(WATER_TEMPERATURE_Update, 6, 100, 100, 0);
SCH_Add_Task(WATER_VALVE_Update, 7, 100, 100, 0);
Listing 14: TTRD03a (system_1769_001-0_c03a.c) [Part 4 of 5]
76
// Add main washer task
SCH_Add_Task(WASHER_CONTROLLER_Update, 0, 1000, 100, 0);
// Add heartbeat task
SCH_Add_Task(HEARTBEAT_Update, 0, 1000, 20, 0);
// Feed the watchdog
WATCHDOG_Update();
break;
}
}
}
/*------------------------------------------------------------------*SYSTEM_Perform_Safe_Shutdown()
Attempt to place the system into a safe state.
Note: Does not return and may (if watchdog is operational) result
in a processor reset, after which the function may be called again.
[The rational for this behaviour is that - after the reset the system MAY be in a better position to enter a safe state.
To avoid the possible reset, adapt the code and feed the WDT
in the loop.]
-*------------------------------------------------------------------*/
void SYSTEM_Perform_Safe_Shutdown(void)
{
// Used for simple error reporting
uint32_t Delay, j;
// Here we simply "fail silent" with rudimentary error reporting
// Other behaviour may make more sense in your design
// *************************************
// NOTE: This function should NOT return
// *************************************
HEARTBEAT_Init();
while(1)
{
// Flicker Heartbeat LED to indicate error
for (Delay = 0; Delay < 200000; Delay++) j *= 3;
HEARTBEAT_Update();
}
}
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 14: TTRD03a (system_1769_001-0_c03a.c) [Part 5 of 5]
77
/*--------------------------------------------------------------------*washer_controller_1769_001-0_c03a.c (Release 2014-11a)
-------------------------------------------------------------------Main (system state) task.
Part of simple initial case study (washing machine controller)
See "ERES" book, Chapter 3.
-*--------------------------------------------------------------------*/
// Project header
#include "../main/main.h"
// Task header
#include "washer_controller_1769_001-0_c03a.h"
// Support functions
#include "../task_support_fns/report_number_7seg_1769_001-0_c03a.h"
// ------ Public variable declarations ----------------------------// Sensors
extern uint32_t
extern uint32_t
extern uint32_t
extern uint32_t
Start_switch_pressed_G;
Selector_dial_reading_G;
Water_level_reading_G;
Water_temperature_reading_G;
// Actuators
extern uint32_t
extern uint32_t
extern uint32_t
extern uint32_t
extern uint32_t
extern uint32_t
extern uint32_t
Drum_motor_required_state_G;
Drum_motor_required_speed_G;
Door_lock_required_state_G;
Detergent_hatch_required_state_G;
Water_heater_required_state_G;
Water_pump_required_state_G;
Water_valve_required_state_G;
// ------ Private data type declarations --------------------------// Possible system states
typedef enum {INIT, START, FILL_DRUM, HEAT_WATER, PRE_WASH,
MAIN_WASH, DRAIN_DRUM, SPIN, FINISHED, FAULT}
eSystem_state;
Listing 15: TTRD03a (washer_controller_1769_001-0_c03a.c) [Part 1 of 8]
78
// ------ Private constants ---------------------------------------// All durations are in seconds (short times here for demo)
#define MAX_FILL_DURATION (100)
#define MAX_WATER_HEAT_DURATION (100)
#define PRE_WASH_DURATION (20)
#define MAIN_WASH_DURATION (20)
#define DRAIN_DRUM_DURATION (10)
#define SPIN_DURATION (10)
// ------ Private variables ---------------------------------------static eSystem_state System_state_G;
static uint32_t Time_in_state_G;
static uint32_t Program_G;
// Ten different programs are supported
// Each one may or may not use detergent
static uint32_t Detergent_G[10] = {1,1,1,0,0,1,0,1,1,0};
// Each one may or may not use hot water
static uint32_t Hot_Water_G[10] = {1,1,1,0,0,1,0,1,1,0};
/*------------------------------------------------------------------*WASHER_CONTROLLER_Init()
Prepare for WASHER_CONTROLLER_Update() task - see below.
-*------------------------------------------------------------------*/
void WASHER_CONTROLLER_Init(void)
{
System_state_G = INIT;
// Used to report current system state in this demo
REPORT_NUMBER_7SEG_Init();
}
Listing 15: TTRD03a (washer_controller_1769_001-0_c03a.c) [Part 2 of 8]
79
/*------------------------------------------------------------------*WASHER_CONTROLLER_Update().
Main (system state) task for the washing-machine controller.
-*------------------------------------------------------------------*/
void WASHER_CONTROLLER_Update(void)
{
// Call once per second
switch (System_state_G)
{
case INIT: // 0
{
// For demo purposes only
REPORT_NUMBER_7SEG_Update((uint32_t) System_state_G);
// Set up initial state
// Motor is off
Drum_motor_required_state_G = STOPPED;
Drum_motor_required_speed_G = 0;
// Pump is off
Water_pump_required_state_G = OFF;
// Heater is off
Water_heater_required_state_G = OFF;
// Valve is closed
Water_valve_required_state_G = CLOSED;
// Detergent hatch is closed
Detergent_hatch_required_state_G = CLOSED;
// Unlock the door
Door_lock_required_state_G = UNLOCKED;
// Wait in this state until START switch is pressed
if (Start_switch_pressed_G == 1)
{
// START *has* been pressed.
// Read the selector dial
Program_G = Selector_dial_reading_G;
// Change state
System_state_G = START;
}
break;
}
Listing 15: TTRD03a (washer_controller_1769_001-0_c03a.c) [Part 3 of 8]
80
case START: // 1
{
// For demo purposes only
REPORT_NUMBER_7SEG_Update((uint32_t) System_state_G);
// Lock the door
Door_lock_required_state_G = LOCKED;
// Start filling the drum
Water_valve_required_state_G = OPEN;
// Release the detergent (if any)
if (Detergent_G[Program_G] == 1)
{
Detergent_hatch_required_state_G = OPEN;
}
// Ready to go to next state
System_state_G = FILL_DRUM;
Time_in_state_G = 0;
break;
}
case FILL_DRUM: // 2
{
// For demo purposes only
REPORT_NUMBER_7SEG_Update((uint32_t) System_state_G);
// Remain in this state until drum is full
if (++Time_in_state_G >= MAX_FILL_DURATION)
{
// Should have filled the drum by now...
System_state_G = FAULT;
}
// Check the water level
if (Water_level_reading_G == 1)
{
// Drum is full
// Stop filling the drum
Water_valve_required_state_G = CLOSED;
// Does the program require hot water?
if (Hot_Water_G[Program_G] == 1)
{
Water_heater_required_state_G = ON;
// Ready to go to next state
System_state_G = HEAT_WATER;
Time_in_state_G = 0;
}
Listing 15: TTRD03a (washer_controller_1769_001-0_c03a.c) [Part 4 of 8]
81
else
{
// Using cold water only
// Ready to go to next state
System_state_G = PRE_WASH;
Time_in_state_G = 0;
}
}
break;
}
case HEAT_WATER: // 3
{
// For demo purposes only
REPORT_NUMBER_7SEG_Update((uint32_t) System_state_G);
// Remain in this state until water is hot
// NOTE: Timeout facility included here
if (++Time_in_state_G >= MAX_WATER_HEAT_DURATION)
{
// Should have warmed the water by now...
System_state_G = FAULT;
}
// Check the water temperature
if (Water_temperature_reading_G == 1)
{
// Water is at required temperature
// Ready to go to next state
System_state_G = PRE_WASH;
Time_in_state_G = 0;
}
break;
}
case PRE_WASH: // 4
{
// For demo purposes only
REPORT_NUMBER_7SEG_Update((uint32_t) System_state_G);
// All wash program involve WASH_01
// Drum is slowly rotated at medium speed
Drum_motor_required_state_G = RUNNING;
Drum_motor_required_speed_G = 50;
if (++Time_in_state_G >= PRE_WASH_DURATION)
{
System_state_G = MAIN_WASH;
Time_in_state_G = 0;
}
break;
}
Listing 15: TTRD03a (washer_controller_1769_001-0_c03a.c) [Part 5 of 8]
82
case MAIN_WASH: // 5
{
// For demo purposes only
REPORT_NUMBER_7SEG_Update((uint32_t) System_state_G);
// Drum is rotated at higher speed
Drum_motor_required_state_G = RUNNING;
Drum_motor_required_speed_G = 70;
if (++Time_in_state_G >= MAIN_WASH_DURATION)
{
System_state_G = DRAIN_DRUM;
Time_in_state_G = 0;
}
break;
}
case DRAIN_DRUM: // 6
{
// For demo purposes only
REPORT_NUMBER_7SEG_Update((uint32_t) System_state_G);
// Pump is activated to drain drum
Water_pump_required_state_G = ON;
// Drum is rotated at low speed
Drum_motor_required_state_G = RUNNING;
Drum_motor_required_speed_G = 10;
// Check the water level
if (Water_level_reading_G == 0)
{
// Drum is empty
// Move to "Spin" state
System_state_G = SPIN;
Time_in_state_G = 0;
}
if (++Time_in_state_G >= DRAIN_DRUM_DURATION)
{
// Drum should have drained by now - fault
System_state_G = FAULT;
Time_in_state_G = 0;
}
break;
}
Listing 15: TTRD03a (washer_controller_1769_001-0_c03a.c) [Part 6 of 8]
83
case SPIN: // 7
{
// For demo purposes only
REPORT_NUMBER_7SEG_Update((uint32_t) System_state_G);
// Drum is rotated at high speed
Drum_motor_required_state_G = RUNNING;
Drum_motor_required_speed_G = 100;
if (++Time_in_state_G >= SPIN_DURATION)
{
System_state_G = FINISHED;
Time_in_state_G = 0;
}
break;
}
case FINISHED: // {
// For demo purposes only
REPORT_NUMBER_7SEG_Update((uint32_t) 10);
// Set up safe state
Drum_motor_required_state_G = STOPPED;
Drum_motor_required_speed_G = 0;
Water_pump_required_state_G = OFF;
Water_heater_required_state_G = OFF;
Water_valve_required_state_G = CLOSED;
Detergent_hatch_required_state_G = CLOSED;
// Wait 10 seconds before unlocking the door
// (to ensure drum stopped, etc)
if (++Time_in_state_G >= 10)
{
// Unlock the door
Door_lock_required_state_G = UNLOCKED;
// Now return to Init state (ready for next wash)
System_state_G = INIT;
Time_in_state_G = 0;
}
break;
}
Listing 15: TTRD03a (washer_controller_1769_001-0_c03a.c) [Part 7 of 8]
84
default:
// Default state
case FAULT: // E
{
// For demo purposes only
REPORT_NUMBER_7SEG_Update((uint32_t) 11);
// Set up safe state
Drum_motor_required_state_G = STOPPED;
Water_pump_required_state_G = OFF;
Water_heater_required_state_G = OFF;
Water_valve_required_state_G = CLOSED;
Detergent_hatch_required_state_G = CLOSED;
// Wait 10 seconds before unlocking the door
// (Ensure drum stopped.)
if (++Time_in_state_G >= 10)
{
// Unlock the door
Door_lock_required_state_G = UNLOCKED;
}
break;
}
}
}
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 15: TTRD03a (washer_controller_1769_001-0_c03a.c) [Part 8 of 8]
85
/*--------------------------------------------------------------------*start_switch_1769_001-0_c03a.c (Release 2014-11a)
-------------------------------------------------------------------Simple switch interface code, with software debounce.
[Reads SW3 on LPCxpresso baseboard.]
-*--------------------------------------------------------------------*/
// Project header
#include "../main/main.h"
// Task header
#include "start_switch_1769_001-0_c03a.h"
// ------ Public variable definitions -----------------------------uint32_t Start_switch_pressed_G = 0;
// ------ Private constants ---------------------------------------// Allows NO or NC switch to be used (or other wiring variations)
#define SW_PRESSED (0)
// SW_THRES must be > 1 for correct debounce behaviour
#define SW_THRES (3)
// ------ Private variable definitions -----------------------------static uint8_t sw3_input = 0;
/*------------------------------------------------------------------*SWITCH_SW3_Init()
Initialisation function for the switch library.
-*------------------------------------------------------------------*/
void SWITCH_SW3_Init(void)
{
// Set up "SW3" as an input pin
// Params: Port : Pin : 1 for o/p, 0 for i/p
GPIO_SetDir(START_SWITCH_PORT, START_SWITCH_PIN, 0);
}
Listing 16: TTRD03a (start_switch_1769_001-0_c03a.c) [Part 1 of 2]
86
/*------------------------------------------------------------------*SWITCH_SW3_Update()
This is the main switch function.
It should usually be scheduled approx. every 10 ms.
-*------------------------------------------------------------------*/
void SWITCH_SW3_Update(void)
{
// Duration of switch press
static uint32_t Duration = 0;
// Read SW3
sw3_input = (GPIO_ReadValue(START_SWITCH_PORT) & START_SWITCH_PIN);
if (sw3_input == SW_PRESSED)
{
Duration += 1;
if (Duration > SW_THRES)
{
Duration = SW_THRES;
Start_switch_pressed_G = 1;
}
else
{
// Switch pressed, but not yet for long enough
Start_switch_pressed_G = 0;
}
}
else
{
// Switch not pressed - reset the count
Duration = 0;
// Update status
Start_switch_pressed_G = 0;
}
}
/*------------------------------------------------------------------*---- END OF FILE -------------------------------------------------*------------------------------------------------------------------*/
Listing 16: TTRD03a (start_switch_1769_001-0_c03a.c) [Part 2 of 2]
87
/*--------------------------------------------------------------------*water_level_1769_001-0_c03a.c (Release 2014-11a)
-------------------------------------------------------------------Placeholder module.
Part of simple initial case study (washing machine controller)
See "ERES" book, Chapter 3.
-*--------------------------------------------------------------------*/
// Project header
#include "../main/main.h"
// Task header
#include "water_level_1769_001-0_c03a.h"
// Fault injection options
#include "../fault_injection/fault_injection_1769_001-0_c03a.h"
// ------ Public variable declarations ------------------------------extern uint32_t Water_pump_required_state_G;
extern uint32_t Water_valve_required_state_G;
// ------ Public variable definitions -------------------------------uint32_t Water_level_reading_G;
/*--------------------------------------------------------------------*WATER_LEVEL_Init()
Prepare for WATER_LEVEL_Update() task - see below.
-*--------------------------------------------------------------------*/
void WATER_LEVEL_Init(void)
{
// Placeholder task
Water_level_reading_G = 0;
}
Listing 17: TTRD03a (water_level_1769_001-0_c03a.c) [Part 1 of 2]
88
/*--------------------------------------------------------------------*WATER_LEVEL_Update().
Placeholder task.
In finished system, this task will read the water level in the
washing-machine drum.
Here we run a simple model.
-*--------------------------------------------------------------------*/
void WATER_LEVEL_Update(void)
{
// Placeholder code
static uint32_t Water_level = 0;
// Is the drum filling?
if ((Water_valve_required_state_G == OPEN) && !WATER_VALVE_FAULT)
{
// Drum is filling
Water_level++;
// Is the drum full of water (ready to wash)?
if (Water_level >= 50)
{
Water_level_reading_G = 1;
}
}
// Is the drum being drained?
if ((Water_pump_required_state_G == ON) && !WATER_PUMP_FAULT)
{
// Drum is being drained
if (Water_level > 0)
{
// Drum is draining
Water_level--;
}
if (Water_level == 0)
{
// Drum is empty
Water_level_reading_G = 0;
}
}
}
/*--------------------------------------------------------------------*---- END OF FILE ---------------------------------------------------*--------------------------------------------------------------------*/
Listing 17: TTRD03a (water_level_1769_001-0_c03a.c) [Part 2 of 2]
89
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement