Tools and Benchmarks for Real-Time Systems

Task sets / Task set generators • Benchmarking crimes

2017-11-18T12:32:32+01:00

Dos and don'ts for benchmarking: http://gernot-heiser.org/benchmarking-crimes.html

Statistics: Posted by Sophie Quinton — Sat Nov 18, 2017

Task sets / Task set generators • Re: SMFF: parameter-driven generation of models of distributed real-time systems

2016-07-19T13:57:07+01:00

Hi Ali,

what are your expectations with SMFF?
If you expect a graphical user interface or something of that kind I am sorry, because there is none.

However, all of the modules come with a example file (that you can compile and run). As everything is open-source you can go ahead and adapt these to your needs and create your own systems.

You can find the examples here:
$SMFF_REPOS/SMFF-Export/SMFF-Export/src/org/tubs/epoc/SMFF/ImportExport/Pdf/Example/Example.java
$SMFF_REPOS/SMFF-PyCPAInterface/SMFF-PyCPAInterface/src/org/tubs/epoc/SMFF/PyCPAInterface/example/AnalysisExample.java
$SMFF_REPOS/SMFF-SystemFactories/SMFF-SystemFactories/src/org/tubs/epoc/SMFF/SystemFactories/Example/Example.java

The slides of the presentation of SMFF can be found here:
http://retis.sssup.it/~tommaso/waters20 ... rchner.pdf
and the Paper describing SMFF and some parameters is here:
http://retis.sssup.it/~tommaso/waters20 ... edings.pdf

Best,
Mischa

Statistics: Posted by Mischa Moestl — Tue Jul 19, 2016

Task sets / Task set generators • Re: SMFF: parameter-driven generation of models of distributed real-time systems

2016-07-14T10:45:19+01:00

Hello,

Thank you Mischa and Pierre for providing help in compiling.

@ Mischa: As you work in the Technische Universität Braunschweig, I though you were the user too. I am sorry, I didn't know that. As the contact links on the SMFF website are broken and both the authors left the university, do you know anybody working in your group whom we can talk to regarding the 2nd question? For me it seams illogical to try compiling and installing, if I don't know how to use (or at least run) the tool.

Best regards,
Ali

Statistics: Posted by syed — Thu Jul 14, 2016

Task sets / Task set generators • Re: SMFF: parameter-driven generation of models of distributed real-time systems

2016-07-13T08:26:17+01:00

Hi, I think SMFF works on my computer :
-try to add the command attribute in eclipse (Runs->Run Configurations->maven build(on the right panel)), and in the Main panel :
add Parameter = log4j.configuration
Value = file:/yourPathTo/SMFF-Core/SMFF-Core/config/src/log4j.xml
see http://stackoverflow.com/questions/2594 ... j-xml-file
Do not forget to add a "Goals", mine is "compile" and it works well.
You can also try to run this in a terminal as :
mvn compile -Dlog4j.configuration=file:/yourPathTo/SMFF-Core/SMFF-Core/config/src/log4j.xml

Best Regards
Pierre

Statistics: Posted by Pierre GONDOIS — Wed Jul 13, 2016

Task sets / Task set generators • Re: SMFF: parameter-driven generation of models of distributed real-time systems

2016-07-12T16:45:19+01:00

Hi Ali and other prospective SMFF users,

let me stress that I am neither a SMFF developer nor the maintainer. So all I can provide you is second hand knowledge. Let me also say that I am not really in depth familiar with the maven build flow used for SMFF; I am not a Java expert either!
But I will try to explain to the best of my knowledge. So in case some Java/Maven experts follow this thread please correct me.

So let me try to answer your questions:

1. How to link the modules Core and SystemFactory?

This is already done by the pom.xml file in the SystemFactory repository. Side note: The pom file is basically the "makefile" of a maven project where you can also specify dependencies of your project such that maven can automatically fetch them from online repositories.

I assumed that Core is in the dependency of SystemFactory, so compiling SystemFactory would suffice, ...

It is correct that SystemFactory specifies SMFF-Core as a dependency. See the pom.xml of the SystemFactory git-SystemFactory for that.

...but it says module org.apache.log4j can not be found. Do I need to grab log4j manually? I am not sure if linking the log4j module will result in successful compilation.

Basically the problem here is that the dependency on log4j is specified as a runtime dependency in the pom.xml of SMFF-Core. Unfortunately SytemFactories need it already at compile time. So for me two solutions to build the project work:

1) remove the scope line in the SMFF-Core pom.xml for log4j:

CODE:

 log4j log4j 1.2.12runtime   <-- REMOVE THIS LINE

2) Add log4j as a build/compile dependency in the pom.xml of SMFF-SystemFactories. Although this seems unreasonable to me since you have SMFF-Core as a dependency anyway. The trick with specifying a dependency in maven, that maven fetches it for you.

After applying one of the two solutions simply navigate to the SMFF-SystemFactories directory and do a:

CODE:

# mvn install

This builds SMFF-SystemFactories and SMFF-Core and installs them in the local maven repository.
If you work with Linux the default maven repository path seems to be $HOME/.m2/repositories/

However, with your second question I can not help you since I am not an SMFF user either. Maybe we have someone else here who can help you with that.
The only wisdom I have for you is that maven installs the SMFF jar files to your local maven repository. If you want to export generated models you should have a look at the SMFF-Export module (or the PyCPA Interface if you want to use the generated Systems there).

Kind regards,
Mischa

Statistics: Posted by Mischa Moestl — Tue Jul 12, 2016

Task sets / Task set generators • Re: SMFF: parameter-driven generation of models of distributed real-time systems

2016-07-12T14:36:58+01:00

Hello,

Thanks for the useful information.

I tried to see how this taskset generator works. Unfortunately, I am not familiar with Maven. I had some questions from Mr. Möstl.

1. How to link the modules Core and SystemFactory? I assumed that Core is in the dependency of SystemFactory, so compiling SystemFactory would suffice, but it says module org.apache.log4j can not be found. Do I need to grab log4j manually? I am not sure if linking the log4j module will result in successful compilation.
2. Once we get it compiled, how do we run it (e.g. command line arguments)? and what is the starting point? Can you please put some examples, perhaps on the actual website?

Hope to hear from you soon,
Best regards,
Ali Syed

Statistics: Posted by syed — Tue Jul 12, 2016

Task sets / Task set generators • Re: SMFF: parameter-driven generation of models of distributed real-time systems

2016-07-07T14:36:55+01:00

Hi,

the git-repository is alive, the description of the URL to the individual modules is just a bit ambiguous.

To clone the four SMFF module repositories in your current directory, the following works:

CODE:

git clone git://smff.git.sourceforge.net/gitroot/smff/SMFF-Coregit clone git://smff.git.sourceforge.net/gitroot/smff/SMFF-SystemFactories git clone git://smff.git.sourceforge.net/gitroot/smff/SMFF-Exportgit clone git://smff.git.sourceforge.net/gitroot/smff/SMFF-PyCPAInterface

Hope that helps!
Mischa

Statistics: Posted by Mischa Moestl — Thu Jul 07, 2016

Task sets / Task set generators • Re: SMFF: parameter-driven generation of models of distributed real-time systems

2016-03-17T12:18:23+01:00

Hello,

The link to git repositories seems to be dead.

Best regards,

Christophe Prévot
PhD student
Member of the SPADES project-team
INRIA Grenoble - Rhône-Alpes

Statistics: Posted by cprevot — Thu Mar 17, 2016

Task sets / Task set generators • Re: a RT task set from Lunar Rover

2016-02-08T11:10:23+01:00

Here is the UPPAAL model of the Lunar Rover control software, its RTOS and its physical environment. We built the model for analyzing a functional-timing problem proposed by the software developers. The model covers a set of tasks which are related to the specific problem. The task set include two types of tasks: periodic and sporadic. Each type of tasks is modeled with a timed-automaton template.

LunarRover.xml

Statistics: Posted by iceslj — Mon Feb 08, 2016

Task sets / Task set generators • Re: a RT task set from Lunar Rover

2016-02-03T15:20:19+01:00

Thanks for the prompt reply, would it be possible for you to post the entire model here?

Thanks,
Sophie

Statistics: Posted by Sophie Quinton — Wed Feb 03, 2016

Task sets / Task set generators • Re: a RT task set from Lunar Rover

2016-02-03T14:22:30+01:00

The paper does not contain the complete UPPAAL model due to the lack of space. But the main parts are in it. The major piece missed in the paper the sporadic task template, which is similar to the periodic task template.

Statistics: Posted by iceslj — Wed Feb 03, 2016

Task sets / Task set generators • Re: a RT task set from Lunar Rover

2016-02-03T14:06:11+01:00

Hi Lijun,

Thanks for the info! Is the complete UPPAAL model available in the paper or are there some parts missing due to lack of space?

Best,
Sophie

Statistics: Posted by Sophie Quinton — Wed Feb 03, 2016

Task sets / Task set generators • a RT task set from Lunar Rover

2016-01-28T17:09:25+01:00

We did a formal verification of Chinese Lunar Rover control software, an embedded real-time multitasking software system running over a home-made real-time operating system (RTOS). The main purpose of the verification is to validate if the system satisfies a time-related functional property. We modeled the RTOS, application tasks and physical environment as timed automata and analyzed the system using statistical model checking (SMC) of UPPAAL. Verification result showed that our model was able to track down undesired behavior in the multitasking system. Moreover, as the modeling framework we designed is general and extensible, it can be a reference method for verifying other real-time multitasking systems.

Link to the paper:
http://link.springer.com/chapter/10.100 ... 06410-9_48

Statistics: Posted by iceslj — Thu Jan 28, 2016

Task sets / Task set generators • Re: TGFF (Task graphs for free)

2015-07-16T10:28:44+01:00

Script

Generate multiple graphs and convert them to our own format (e.g. XML)

The script assumes that there is no seed command in the input tgffopt file.

CODE:

#!/bin/bash# --------------------------------------------------------- Functions -------------------------------------------------------------------function printUsage {echo "Usage: $0      "echo "1) Generates  tasksets in  from  using the initial file as  and seeds them from 0 to ."echo "2) Converts generated tasksets to desired format using "}# --------------------------------------------------- Check command line arguments -------------------------------------------------------if [ ! $# == 6 ]thenprintUsageexitfiTGFF=$1TGFFConverter=$2START_FILE=$3NUMBER_OF_TASKSETS=$4CORES_REQUIRED=$5OUTPUT_FOLDER=$6if [ ! -e "$TGFF" ]thenecho Unable to find TGFFexit 2elif [ ! -e "$TGFFConverter" ]thenecho Unable to find $TGFFConverterexit 2elif [ ! -e "$START_FILE" ]thenecho Unable to find $START_FILEexit 2elif [ $NUMBER_OF_TASKSETS -eq 0 ]thenecho Please define valid number of tasksets and coresexit 2fi# Remove previous tasksets (if exist)rm -rf $OUTPUT_FOLDER# Create foldermkdir $OUTPUT_FOLDERfor (( c=0; c

Feel free to modify.Statistics: Posted by syed — Thu Jul 16, 2015



Task sets / Task set generators • TGFF (Task graphs for free)
2015-07-16T10:53:11+01:00

Task Graphs For Free (TGFF)

Documentation and published papers can be found on the link provided above.


Short summary:

Generate reproducible task graphs with homo/heterogeneous processors, with/without communication messages and much more.


Pros:
Easy to understand/use.
Reproducible task sets.
Wide range of parameters can be generated.

Cons: 
No direct control over utilization.
Utilization distribution: Only Gaussian. (not written in the documentation, but found experimentally)
No support for conditional checks between same parameters (e.g. Sum of Utilization <= 1) or multiple parameters (e.g. Sum of WCET of longest chain <= deadline of that chain)

Usage:

Each task defines a type (unique or not) instead of all possible parameters (WCET, criticality levels, etc.). Once the type table is generated, corresponding tables for parameters (WCET, criticality levels, etc.) can be generated.

Sample input file:CODE: 
seed 1 # Specifies uniqueness of the taskset. Reproduce using same seed.tg_cnt 1 # Number of task graphstask_cnt 20 5 # Number of tasks: Average, +- Rangetask_degree 2 2 # Predecssors and successors per tasktask_trans_time 200 # Hyperperiod Controlperiod_mul 0.1, 0.5, 1, 2 # Multiples for 'task_trans_time' to be used as Periodsperiod_g_deadline 1 # Implicit deadlinestask_unique 1 # Task types become uniquetg_writeeps_writetable_label PROC # New table called PROCtable_cnt 1 # Types of tables (i.e. processors, homogeneous in this case)table_attrib # Number of values per tabletype_attrib  WCET 50 40 # 'WCET': Average, +- Rangepe_write # Create processor tabletable_label COMMUNtable_cnt 1table_attribtype_attrib  Mess_Len 5 4 # Message length: Average, +-Rangetrans_write # Create message table
Statistics: Posted by syed — Thu Jul 16, 2015