1. The Fixed Logical Execution Time (FLET) Assumption Tom Henzinger University of California, Berkeley

2. The History of Computer Science: Lifting the Level of Abstraction The “assembly age”: Programming to the platform High-level languages: Programming to the application Compilation: perhaps “the” success story of computer science It is feasible to abstract the platform.

3. The History of Computer Science: Lifting the Level of Abstraction The “assembly age”: Programming to the platform High-level languages: Programming to the application Automatic program synthesis: No more programming Compilation: perhaps “the” success story of computer science Code generation from specifications: still mostly a dream It is feasible to abstract the platform. It is not feasible to abstract algorithms and data structures.

4. Current Practice in Embedded Software Design Some manual programming to the platform Some automatic code generation from models -often inefficient -often unpredictable -e.g. Real-Time Workshop -difficult to reuse -difficult to verify -requires systems experts

5. Current Practice in Embedded Software Design Efficient code (scheduled by RTOS) Mathematical models (e.g. Simulink) Code verification difficult Code generation difficult

6. Advocated Practice in Embedded Software Design Efficient code (possibly schedule-carrying) Mathematical models (e.g. Simulink) The missing link: platform-independent programming model

7. Efficient code (possibly schedule-carrying) Mathematical models (e.g. Simulink) The missing link: platform-independent programming model Verification Compilation Advocated Practice in Embedded Software Design

8. Efficient code (possibly schedule-carrying) Mathematical models (e.g. Simulink) The missing link: platform-independent programming model Verification Compilation -verifiable -reusable -efficiently implementable Advocated Practice in Embedded Software Design

9. Efficient code (possibly schedule-carrying) Mathematical models (e.g. Simulink) The missing link: platform-independent programming model Verification Compilation -verifiable -reusable -efficiently implementable Reusable = composable + portable Advocated Practice in Embedded Software Design

10. Platform P Requirement A Program A Advocated Practice in Embedded Software Design Requirement B Program B + + Compositionality

11. Platform P Requirement A Program A Advocated Practice in Embedded Software Design Requirement B' Program B' + + Compositionality

12. Requirement A Platform P Advocated Practice in Embedded Software Design Platform Q Portability Program A

13. Efficient code (possibly schedule-carrying) Mathematical models (e.g. Simulink) The missing link: platform-independent programming model Verification Compilation -concurrency -environment time -distribution -platform (CPU) time Advocated Practice in Embedded Software Design

14. Efficient code (possibly schedule-carrying) Mathematical models (e.g. Simulink) The missing link: platform-independent programming model Verification Compilation -concurrency -environment time -distribution -platform (CPU) time -concurrency -environment time Advocated Practice in Embedded Software Design

15. Efficient code (possibly schedule-carrying) Mathematical models (e.g. Simulink) The missing link: platform-independent programming model Verification Compilation -descriptive -mathematics / logic -prescriptive -algorithms + data structures -concurrency -environment time Advocated Practice in Embedded Software Design

16. Efficient code (possibly schedule-carrying) Mathematical models (e.g. Simulink) The missing link: platform-independent programming model Verification Compilation -descriptive -mathematics -prescriptive -algorithms + data structures -concurrency -environment time -prescriptive -virtual machine Advocated Practice in Embedded Software Design

17. Efficient code (possibly schedule-carrying) Mathematical models (e.g. Simulink) The missing link: platform-independent programming model Verification Compilation Advocated Practice in Embedded Software Design e.g. What is the control equation? What is the sampling rate? e.g. Which procedure computes the control equation? Which event triggers the computation? e.g. Which CPU executes the control procedure? What priority has the execution?

18. Efficient code (scheduled by RTOS) Mathematical models (e.g. Simulink) The missing link: platform-independent programming model First Attempt Priorities

19. Efficient code (scheduled by RTOS) Mathematical models (e.g. Simulink) Programming model: Priorities First Attempt Not sufficiently abstract -not about environment time -not compositional

20. Efficient code (scheduled by RTOS) Mathematical models (e.g. Simulink) Programming model: Priorities First Attempt Not sufficiently abstract -not about environment time -not compositional Scheduling theory Correctness? Efficiency!

21. Efficient code (scheduled by RTOS) Mathematical models (e.g. Simulink) The missing link: platform-independent programming model Second Attempt Synchrony assumption: platform time infinitely faster than environment time

22. Efficient code (scheduled by RTOS) Mathematical models (e.g. Simulink) Synchronous programming languages (Esterel, Lustre, Signal, etc.) Second Attempt Too abstract difficult to compile to resource-constrained, distributed platforms

23. Efficient code (scheduled by RTOS) Mathematical models (e.g. Simulink) Synchronous programming languages (Esterel, Lustre, Signal, etc.) Second Attempt Too abstract difficult to compile to resource-constrained, distributed platforms Correctness! Efficiency?

24. Efficient code (possibly schedule-carrying) Mathematical models (e.g. Simulink) The missing link: platform-independent programming model Our Attempt FLET assumption: -verifiable -efficiently implementable -composable -portable

25. The FLET (Fixed Logical Execution Time) Assumption Software Task read sensor input at time t write actuator output at time t+d, for fixed d

26. Software Task read sensor input at time t write actuator output at time t+d, for fixed d d>0 is the task's "logical execution time" The FLET (Fixed Logical Execution Time) Assumption

27. The programmer specifies d (could be any event) to solve the problem at hand. The compiler ensures that d is met on a given platform (hardware resources and performance); otherwise it rejects the program. The FLET Programming Model

28. time ttime t+d possible physical execution on CPU buffer output The FLET (Fixed Logical Execution Time) Assumption

29. 50% CPU speedup Portability

30. Task 2 Task 1 Composability

31. -timing predictability: minimal jitter -function predictability: no race conditions Verifiability through Predictability (Internal Determinism)

32. output as soon as ready Contrast FLET with Standard Practice

33. Race Contrast FLET with Standard Practice

34. … yes but, what about the sacrifice in performance ?!

35. Test Case: Flight Control Software UC Berkeley (Horowitz, Liebman, Ma, Koo, Sangiovanni-Vincentelli, Sastry). Two connected CPUs.

36. Flight Control Software Architecture

37. 200 Hz 400 Hz 200 Hz 1 kHz Flight Control Software Architecture

38. 1. Concurrent periodic tasks: -sensing -control law computation -actuating 2. Multiple modes of operation: -navigational modes (autopilot, manual, etc.) -maneuver modes (taxi, takeoff, cruise, etc.) -degraded modes (sensor, actuator, CPU failures) Platform-independent Software Model

39. Mode 1 Mode 4Mode 3 Mode 2 Task S 400 Hz Task C 200 Hz Task A 1 kHz Task S 400 Hz Task C 200 Hz Task A’ 1 kHz Task C’ 100 Hz Task A 1 kHz Task S 400 Hz Task C 200 Hz Task A 2 kHz Task A” 1 kHz Condition 1.2 Condition 2.1 Platform-independent Software Model

40. Host code e.g. C Glue code e.g. Giotto Functionality. -Environment time, not platform time. -Concurrency, not distribution. Platform-independent Software Model Timing and interaction. This kind of software is understood: Host code may (sometimes) be generated automatically. The software complexity lies in the glue code (minimize jitter!): Giotto enables requirements-driven rather than platform-driven glue-code programming. -No time. -Sequential.

41. 1. The Giotto Programmer’s Model: Time-triggered FLET 2. The Giotto Compiler

42. The Giotto Programmer’s Model Programming in terms of environment time: Programmer’s fiction: -time-triggered task invocation -tasks are functions with a fixed duration -platform offers sufficient performance Implementation in terms of platform time: Compiler must maintain programmer’s fiction: -needs access to global time, no other platform requirements -tasks may finish early, but outputs cannot be observed early -tasks may be preempted and distributed

43. 1. Units of scheduled host code (application-level tasks). e.g. control law computation 2. Units of synchronous host code (system-level drivers). e.g. device drivers Task Input portsOutput ports Task Task driver loads task input ports. Functional Components

44. Task Driver Input ports loaded. Driver execution in environment time 0. Task execution in environment time d. Output ports read. Sensor/output ports read. Sensor Actuator Actuator/input ports loaded. Time t Time t+d d Task duration Environment Timeline (defined by Giotto semantics)

45. Task Driver Input ports loaded. Output ports read. Sensor Time t Time t+d d Task on CPU. Actuator Platform Timeline (chosen by Giotto compiler)

46. The Giotto compiler chooses for a given platform a platform timeline that is value equivalent to the environment timeline defined by the Giotto semantics. Internal Determinism: For a given sequence of sensor readings, the corresponding sequence of actuator settings is uniquely determined (i.e., there are no race conditions). Platform Independence ensures Predictability

47. Navigation Control Simplified Helicopter Software Sensors Actuators i s a 10 5

48. Navigation Control Simplified Helicopter Software Sensors Actuators i s a 10 5 Simulink / legacy design

49. … mode Flight ( ) period 10ms { actfreq 1 do Actuator ( actuating ) ; taskfreq 1 do Control ( input ) ; taskfreq 2 do Navigation ( sensing ) ; } … Helicopter Software: Giotto Syntax Navigation Control i s a 10 5

50. aia t+10ms s Task Navigation Control t+10ms tt t+5ms i ss Helicopter Software: Environment Timeline Block of synchronous code (nonpreemptable) Scheduled tasks (preemptable)

51. t+10ms Task t+10ms tt t+5ms Single-CPU Helicopter: Platform Timeline (EDF)

52. t+10ms ttt+5ms t+7ms HeliCtr HeliNav TDMA Slot HeliNet Two-CPU Helicopter: Platform Timeline (Time-triggered Communication)

53. t+10mstt t+5ms HeliCtr HeliNav Message HeliNet Two-CPU Helicopter: Platform Timeline (Event-triggered Communication) t+10ms

54. 1. The Giotto Programmer’s Model 2. The Giotto Compiler

55. The Giotto Compiler Giotto Program Native Code Tasks and Drivers Giotto-P Platform Specification -topology (CPUs, networks) -performance (WCETs, latencies) -APIs (RTOSs, protocols) Executables Giotto Compiler Functionality Timing Interaction Platform “Failure” either Giotto-P overconstrained, or compiler not smart enough (distributed scheduling problem) or

56. Closing the Gap: Annotated Giotto Giotto Program Native Code Tasks and Drivers Giotto-P -topology (CPUs, networks) -performance (WCETs, latencies) -APIs (RTOSs, protocols) Executables Giotto Compiler Functionality Timing Interaction Platform “Failure” either Giotto-PM overconstrained, or compiler not smart enough (global scheduling problem) or Giotto-PM -assign tasks to CPUs -assign connections to networks Map

57. Closing the Gap: Annotated Giotto Giotto Program Native Code Tasks and Drivers Giotto-P Executables Giotto Compiler Functionality Timing Interaction Platform “Failure” Giotto-PMC overconstrained (local scheduling problems solvable) or Giotto-PM Map Giotto-PMC Communication -assign connections to TDMA slots (say) -topology (CPUs, networks) -performance (WCETs, latencies) -APIs (RTOSs, protocols) -assign tasks to CPUs -assign connections to networks

58. Code Generation Giotto Program Native Code Tasks and Drivers Giotto-P Giotto Compiler Functionality Timing Interaction Platform Giotto-PM Map Giotto-PMC Communication -assign connections to TDMA slots (say) -topology (CPUs, networks) -performance (WCETs, latencies) -APIs (RTOSs, protocols) -assign tasks to CPUs -assign connections to networks or “Failure” VxWorksOSEK …

59. Code Generation: The Embedded Machine Giotto Program Native Code Tasks and Drivers Giotto-P Embedded Machine code Giotto Compiler Functionality Timing Interaction Platform or Giotto-PM Map Giotto-PMC Communication -assign connections to TDMA slots (say) -topology (CPUs, networks) -performance (WCETs, latencies) -APIs (RTOSs, protocols) -assign tasks to CPUs -assign connections to networks “Failure” Embedded Machine interpreter

60. The Embedded Machine -a virtual machine that mediates the interaction of physical processes (sensors and actuators) and software processes (tasks and drivers) in real time -the Giotto compiler can be retargeted to a new platform by porting the Embedded Machine

61. Environment Software Software Processes: platform time Environment Processes: environment time The Embedded Machine

62. Environment Software Schedulability: time safety checking for platform time Reactivity: programming in environment time The Embedded Machine: Time is like Memory Embedded Machine

63. Environment Ports Task Ports Driver Ports Embedded Machine task triggers environment triggers sense actuate read write call drivers The Embedded Machine schedule tasks e.g. clock e.g. task completion

64. Enable trigger: future( g,B:) B: Schedule task: schedule( T ) T Call driver: call( d ) d g The Embedded Machine: Three Instructions Execute driver d now. Hand task t over to the system scheduler (RTOS). Have code B executed as soon as trigger g becomes true.

65. ai a t+10ms s Task Navigation Control t+10ms ttt+5ms i ss B1:call( actuate ) call( sense ) call( input ) schedule( Control ) schedule( Navigation ) future( now+5, B2:) relax Giotto Code Generation

66. ai a t+10ms s Task Navigation Control t+10ms tt t+5ms i ss B2:call( sense ) schedule( Navigation ) future( now+5, B1:) relax Giotto Code Generation

67. Embedded Machine Code combines Synchronous code: -Embedded Machine instructions and drivers -kernel context (trigger-related interrupts disabled) Scheduled code: -tasks -user context (trigger-related interrupts enabled)

68. Real-Time Programs Task-triggered code: -triggers on environment and task ports -observable behavior depends on environment and scheduler Environment-triggered code: -triggers only on environment ports -observable behavior depends on environment only All Giotto programs result in environment-triggered code.

69. Real-Time Programs Time-safe code: No driver accesses a scheduled task before completion. Time-live code: There is no infinite synchronous computation. All Giotto programs result in time-live code. Time safety is checked by the Giotto compiler. Alternatively, time safety violations can be can be handled through run-time exceptions.

70. Real-Time Programs Time Safety: Violations due to combination of -insufficient synchrony (environment events too frequent) -insufficient platform utilization (scheduler too weak) -insufficient platform performance (hardware too slow) Our approach systematically integrates synchronous programming and scheduling theory.

71. Distributed Giotto Compiler Giotto Compiler Giotto Program E Code

72. Distributed Giotto Compiler Giotto Compiler with Time Safety Checker Giotto Program E Code Giotto-P Platform Spec Proof of Time Safety = Schedule

73. Schedule-Carrying Code Giotto Compiler with Time Safety Checker Giotto Program E Code Giotto-P Platform Spec Proof of Time Safety (Schedule) +

74. Schedule-Carrying Code Giotto Compiler with Time Safety Checker Giotto Program E Code Giotto-P Platform Spec S Code (Executable Schedule) +

75. ai a t+10ms s Task Navigation Control t+10ms ttt+5ms i ss B1:call( actuate ) call( sense ) call( input ) schedule( Control ) schedule( Navigation ) future( now+5, B2:) relax Giotto Code Generation

76. ai a t+10ms s Task Navigation Control t+10ms ttt+5ms i ss B1:call( actuate ) call( sense ) call( input ) schedule( Control ) schedule( Navigation ) future( now+5, B2:) dispatch( Navigation, now+5 ) dispatch( Control, now+5 ) idle( now+5 ) Giotto EDF Schedule Generation

77. Advantages of SCC (Schedule-Carrying Code) Advantages of PCC: -schedule generation is harder than schedule validation (nonpreemptive or distributed case: NP-hard vs. linear) -once generated, schedule can be reused -compiler does not need to be trusted

78. Advantages of SCC (Schedule-Carrying Code) Advantages of PCC: -schedule generation is harder than schedule validation (nonpreemptive or distributed case: NP-hard vs. linear) -once generated, schedule can be reused -compiler does not need to be trusted Advantages of Executability: -flexible scheduling scheme -no system scheduler (RTOS) necessary -overhead reduction of up to 80%

79. The Giotto Project www.eecs.berkeley.edu/~tah/giotto Software Tools: Simulink to Giotto translator (Pree) Giotto to E code compiler (Kirsch) E Machine on Linux, VxWorks, OSEK (Kirsch) S code time safety checker (Matic) Integrated E and S machine on StrongARM (Kirsch, Sanvido) Event-triggered xGiotto (Ghosal, Sanvido) Applications: Lego Mindstorms (Horowitz, Kirsch, Majumdar) Zurich helicopter (Kirsch, Sanvido) Berkeley helicopter (Horowitz, Liebman, Ma, Sastry) BMW electronic throttle control (Kirsch, Pree)

80. GIOTTO: FLET for periodic tasks with time-triggered mode switching Mode 1 Mode 2 Task S Task C Condition 1.2

81. xGIOTTO: General (event-triggered) FLET programming 1.Schedule Instruction: schedule Task by Event ; 2.Reaction Block: react { when Event do Block ; whenever Event do Block ; begin … end ; } until Event ; logical deadline

82. xGIOTTO: General (event-triggered) FLET programming If all events can happen at any time, then few programs would be time-safe. However, nested reaction blocks can be used for event scoping (i.e., not all events are listened to all the time).

83. xGIOTTO: General (event-triggered) FLET programming If all events can happen at any time, then few programs would be time-safe. However, nested reaction blocks can be used for event scoping (i.e., not all events are listened to all the time). xGiotto = Structured Real-Time Programming

84. The Giotto Project www.eecs.berkeley.edu/~tah/giotto Participants: Ben Horowitz Arkadeb Ghosal Christoph Kirsch Slobodan Matic Marco Sanvido Sponsors: DARPA SEC (Software-Enabled Control) NSF CHESS (Center for Hybrid & Embedded Software Systems)