I successfully defended and wrapped up my PhD! Yaaaay!

Big thanks goes to my advisor Alan Hu, committee members Mark Greenstreet and Ken McMillan, university examiners Gail Murphy and Karthik Pattabiraman, as well as my external examiner Ranjit Jhala. The final version of my PhD thesis is available online and can be found here.

Soon after I wrapped up my PhD, I joined Carnegie Mellon University in Silicon Valley as a postdoctoral fellow. Here I am working on a NASA-funded project and collaborating with Dimitra Giannakopoulou and Vishwanath Raman. The goal of the project is improving coverage of testing and checking of NASA’s air traffic control system using compositional techniques. I am sure that lots of challenging and exciting work is ahead of me!

I just learned that Boogie recently switched to using .NET 4, and therefore the tutorial on how to run Boogie under Linux using .NET 2 I wrote a couple of months ago is now deprecated. I played a little bit with the latest version of Boogie under Linux and managed to get it to work, this time using Mono.

So, first you have to get a Linux binary of Z3. You can always download the latest SMT Competition Z3 Linux binary from the Z3′s download page (look for Other platforms). Copy the bin/z3 executable into Boogie’s folder and rename it to z3.exe.

Alternatively, you can try to get the Windows Z3 binary to work under Linux using wine. So, to be able to run z3.exe, install wine (I am using wine-1.1.28) and download the winetricks script (I am using winetricks version 20090815). Then, using winetricks install vcrun2008

winetricks vcrun2008

Now try typing

wine z3.exe /h

to see if Z3 is working. For Boogie to be able to call Z3, you still have to achieve that z3.exe can be executed without the wine command in front. This is the tricky part that is accomplished using binfmt_misc and doing the following:

su
mount binfmt_misc -t binfmt_misc /proc/sys/fs/binfmt_misc
echo ':DOSWin:M::MZ::/usr/bin/wine:' > /proc/sys/fs/binfmt_misc/register

Try calling Z3 without wine

z3.exe /h

to see if Z3 is still working.

Second, install the latest version of Mono that supports .NET 4. Boogie is going to fail if the environment variable ProgramFiles is not set, but you can fool it by setting ProgramFiles to essentially anything:

export ProgramFiles=/home/

Create a simple bpl file test.bpl and check if just Boogie without Z3 is working using Mono:

mono --runtime=v4.0.30319 Boogie.exe test.bpl /noVerify

Finally, check if Boogie/Z3 combination is working together and execute:

mono --runtime=v4.0.30319 Boogie.exe test.bpl

You can also get rid of the ugly –runtime switch, if you edit Boogie.exe.config like this:

<?xml version="1.0"?>
<configuration>
  <startup>
    <supportedRuntime version="v4.0.30319" sku=".NETFramework,Version=v4.0"/>
  </startup>
</configuration>

And voila, you are done!
I would like to thank Vladimir Klebanov for his valuable feedback and help, and any additional feedback, comments, or improvements would be appreciated.

Domagoj and I recently added some new material to our Guidebook for Graduate Studies Abroad and this 2nd extended online edition is now available for download.

Word cloud summary of the POPL 2011 paper.

The final, camera-ready version of the delay-bounded scheduling paper is now available and can be obtained here. Inspired by others, I also started playing with word clouds a little bit, and on the right you can find a nice word cloud summary of this paper. Enjoy!

The paper titled “Delay-Bounded Scheduling” I coauthored with Michael Emmi and Shaz Qadeer got accepted to the 38th ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages (POPL 2011). The camera-ready version will be ready and available in a couple of weeks. In the meantime, those of you who are eager to learn what delay-bounded scheduling is all about can check out our tech-report.

Software verification tools that I have built in the last couple of years use the combination of Boogie verification condition generator and Z3 SMT solver as their back end. Both Boogie and Z3 are developed by Microsoft Research, and therefore, understandably, Windows is their platform of choice. However, I’ve often had a need to use Boogie and Z3 on Linux, since some of my tools (e.g. SMACK) are Linux based. Next, I am going to describe how I managed to pull that off (disclaimer: there are probably better ways of doing this, but this one worked for me and that’s good enough)…

I am using openSuse 11.2 (Linux version 2.6.31.12), although I am sure this would work on other Linux distributions/versions as well. Before we start, you will, of course, also need Boogie and Z3 binaries, which you can get from the respective websites. Make sure you copy z3.exe into Boogie’s bin folder and run Boogie under Windows to see if it works.

Let’s go into Linux now. To be able to run z3.exe, install wine (I am using wine-1.1.28) and download the winetricks script (I am using winetricks version 20090815). Then, using winetricks install vcrun2008

winetricks vcrun2008

Now try typing

wine z3.exe /h

to see if Z3 is working.

For Boogie you will have to install support for .NET (i.e. dotnet20), again using winetricks

winetricks dotnet20

See if Boogie is working by typing

wine Boogie.exe

Note that I do get some weird error messages (e.g. fixme:shell:URL_ParseUrl failed to parse L”Core”), but those (surprisingly) don’t seem to matter, everything still works.
Finally, check if Boogie/Z3 combination is working together and execute

wine Boogie.exe some_simple_bpl_file.bpl

And voila, you are done!

Note that I tried many other combinations as well (e.g. mono + wine), but this one seems to work the best. Any feedback would be appreciated, especially if you figure out how to get rid of those annoying error messages .

The work I did with Naghmeh and Alan on empirical evaluation of context-bounded translations for concurrent software was accepted at SPIN 2010. We investigate source-to-source translations: a concurrent program, given a bound on the number of context switches, is translated into a corresponding sequential program. In the paper, we compare three different translations in the verification-condition-checking (using SMT solvers) paradigm. We evaluate their scalability under a wide range of experimental conditions: we draw interesting conclusions and provide practical guidance for using context-bounded analysis in a VC-checking paradigm.

The final version of this paper is available here.

And here is what Shaz Qadeer said about the paper: “I really like the paper, quite independent of its focus on evaluation of context-bounded translations. There are very nice insights in the paper about the difference between finite-state symbolic model checking and the VC-based techniques. In fact, I would recommend this paper to anybody who has a model checking background and is beginning to use theorem provers.” Thanks a lot Shaz!

I am excited to announce that I’ll be joining Cadence Research Laboratories in fall! I am going to continue doing research on effective, automatic techniques for analysis, verification, and testing of software. Most likely I’ll be moving to the Bay Area in October.

Woohoo! I won silver medal in the ACM Student Research Competition at ICSE 2010 with my competition entry titled “STORM: Static Unit Checking of Concurrent Programs”. It was a tough competition, with Code Bubbles winning gold and Thomas Fritz, also from UBC, taking bronze. Furthermore, this is not the end of the story for STORM since top three winners are invited to compete in the ACM Student Research Competition Grand Finals.

Communications of the ACM (CACM) recently published an article titled “A Few Billion Lines of Code Later: Using Static Analysis to Find Bugs in the Real World”. Written by the Coverity crowd, the article talks about their experience with commercializing a static bug-finding tool. The authors list practical problems they run into that you typically wouldn’t see when building a research prototype, and also describe how they solved (or worked around) them. Interesting and entertaining…