[klee-dev] rationale behind the parameters used in the KLEE OSDI paper

[Lei Zhang]

Hi Cristian,

Just another question about the KLEE paper. In footnote 7 you said that "We
ran the test suite using the commands: env RUN_EXPENSIVE_TESTS=YES
RUN_VERY_EXPENSIVE_TESTS=YES make check and make check-root (as root). A
small number of tests (14 out of 393) which require special configuration
were not run; from manual inspection we do not expect these to have a
significant impact on our results." How did you calculated the number of
393? We run this command too, but the number reported by "make check" and
"make check-root" is totally about 150. Besides, we have also tried using
Linux audit <http://linux.die.net/man/8/auditd> to record every invocation
of the compiled binaries. The total number of invocations is 8056. In order
to guarantee our evaluation's sanity, we need your help about this
discrepancy. Could you tell us how the number of 393 is calculated? Thank
you in advance!

Best regards,

On Mon, Feb 11, 2013 at 1:12 PM, Cristian Cadar <c.cadar at imperial.ac.uk>wrote:

> On 08/02/13 14:44, Lei Zhang wrote:
> > So this means before replay a ktest, I should remove the old sanbox
> > directory and uncompress sandbox.tgz to create a new one, and then
> > copy the native utility compiled with gcov support to the new sandbox
> > directory and then reply the ktest?
> Hi Lei, yes, this is essentially correct; however, to better mimic the
> symbolic execution environment, what we did was to invoke klee-replay from
> within the sandbox directory, but without copying the native executable in
> the sandbox directory.  Finally, don't forget to reset the environment
> variables before you replay a test.
>
> Best wishes,
>
> Cristian
>
> ______________________________**_________________
> klee-dev mailing list
> klee-dev at imperial.ac.uk
> https://mailman.ic.ac.uk/**mailman/listinfo/klee-dev<https://mailman.ic.ac.uk/mailman/listinfo/klee-dev>
>



-- 
Lei Zhang,
Department of Computer Science & Technology,
School of Electronics Engineering and Computer Science,
Peking University,
No.5 Yiheyuan Road Haidian District,
Beijing, 100871, P.R.China
-------------- next part --------------
HTML attachment scrubbed and removed