Integrating Embedded System Design Tools in the Generic Modeling Environment

(1)

1 of 141

Integrating Embedded System Design Tools in the Generic Modeling Environment

Peter Tureby, Fredrik Wester

Final Project at the Embedded Systems Lab Linköping University, Sweden

(2)

Work Packages

 WP1: SymTA/S model

 2005-04-30

 WP2: Interface to holistic analysis

 2005-04-30

 WP3: Worst-case scenario visualization

 2005-04-30: Can it be done in GME?

 WP4: Complex modeling

 Model to be received from Paul: 2005-05-09

 2005-06-06

 WP5: Interface to Traian’s analysis tool

 2005-06-17

 WP6: Documentation

 Start latest on 2005-06-13

(3)

3 of 143

Work Package #1

 SymTA/S model

 Implement the SymTA/S model in GME

 Same entities and attributes

 Specify the architecture separately

 What are the constraints and how to write them?

 Examples

 Model Kai Richter’s RTSS example

 Model a more complex example

(4)

WP1 Model: Process

 Process

 Name (process name, or “identifier”)

 BCET (best case execution time on the resource it is mapped to, in milliseconds)

 WCET (worst case execution time on the resource it is mapped to, in milliseconds)

 Priority (process priority, natural number)

 Period (how often is the process executed, natural number)

 Jitter (optional; natural number)

 Offset (optional; natural number)

 Deadline (optional; natural number)

 Note

 See my thesis (page 144) for a definition of these terms

(5)

5 of 145

WP1 Model: Message and Resource

 Message

 Name

 Size (the size of the message, in bits)

 Resource

 Name

(6)

WP1 Model: Notes

 Notes:

 It is assumed that all processes and messages are scheduled using

“Fixed-Priority Preemptive Scheduling” (FPS)

 Processes can be connected to each other in an application graph

 Processes can be connected through messages…

 For example, P1->m1->P2

 …but also directly

 For example, P1->P2

 The application graph should have no cycles

 Resources can be connected to each other

 For example, several CPUs are connected to a bus

(7)

7 of 147

Work Package #2

 Interface to holistic analysis

 Holistic analysis implementation from MAST

 Decide which implementation to use

 Traian’s (C)

 Sorin’s (C)

 Integrate the analysis with WP1

 How to pass the model to the analysis tool?

 How to send the results back to GME?

 Analyze the two examples from WP1

(8)

Work Package #3

 Worst-case scenario visualization

 For a given process, visualize it’s worst-case scenario

 How should a worst-case scenario be depicted?

 See SymTA/S for an example

 Is it possible to use GME for the visualization, or do you have to build an external tool?

 Write by hand the worst-case scenario of a process in one of the examples from WP1

 The worst-case scenario is produced by the analysis tool

 How to pass it to GME?

(9)

9 of 149

Work Package #4

 Complex modeling

 Decide on a more complex model (responsible:

Paul, Sorin, Traian)

 Extend the SymTA/S model to the complex model

 What are the constraints and how to write them?

 Three examples

 Model an example from Traian

 Hierarchical scheduling and FlexRay

 Model an example from Sorin

 Stochastic execution times

 Model an example from Paul

 Multi-cluster

(10)

Work Package #5

 Interface to Traian’s analysis tool

 Integrate Traian’s analysis tool in GME

 Complex modeling

 Traian’s example from WP4

(11)

11 of 1411

Work Package #6

 Documentation

 The metamodels and the source code should be documented in English

 Report

 Preferably in English (Swedish is OK)

 Preferably written in LaTeX (Word, or FrameMaker is also fine)

 Presentation

 In English

 Presented in an ESLAB meeting

 Before the summer?

 Use this template in PowerPoint

(12)

Action Points

 Discuss with Paul

 Clarify what has to be done for each WP

 Decide who does what work package

 Decide on milestones

(13)

13 of 1413

Report ToC

 Suggested Table of Contents

 Introduction

 Embedded systems design

 Modeling

 Analysis

 Motivation

 Analysis tools at ESLAB

 Related work

 SymTA/S interface

 Aires’ interface in GME

 MAST interface

 Modeling using GME

 Generic Modeling Environment

 Simple (SymTA/S) model

 Examples

 Complex model

 Examples

 Tool integration in GME

 Design choices

 ESLAB’s tools integration

 Examples

 Worst-case scenario visualization

 Examples

 Conclusions and future work

 References